TWI672603B - Grouping method for asthma patients according to different genders - Google Patents

Abstract

The invention relates to a grouping method for asthma patients according to different genders, which comprises: A. Establishing a group mode database subsystem for storing a plurality of asthma grouping modes of different genders; B. providing a detection a subsystem for detecting a plurality of physiological factors of a group of asthma patients; C. providing a calculation module for calculating physiological factors of the group of asthma patients to be grouped; and D. providing an output module for using The clustering result of the patient to be divided into asthma is output.

Description

A grouping method for asthma patients based on different genders

The invention relates to a grouping method for asthma patients according to different genders, which comprises: A. Establishing a group mode database subsystem for storing a plurality of asthma grouping modes of different genders; B. providing a detection a subsystem for detecting a plurality of physiological factors of a group of asthma patients; C. providing a calculation module for calculating physiological factors of the group of asthma patients to be grouped; and D. providing an output module for using The clustering result of the patient to be divided into asthma is output.

Asthma is a common chronic airway inflammatory disease with variable airflow obstruction affecting all age groups worldwide. Asthma is usually associated with airway inflammation, overreaction and stenosis, with clinical symptoms such as wheezing, coughing shortness and chest tightness, as well as rhinitis, sinusitis, nasal polyps and atopic dermatitis. According to the World Health Organization (WHO), nearly 235 million people currently have asthma, and in the past few years, the prevalence rate in most parts of the world has temporarily increased or remained stable. In Taiwan, epidemiological studies using the ISAAC questionnaire showed that the prevalence of asthma has the same trend, from 4.5% in 1995 to 6.0% in 2001, while another study shows asthma severity and merger in elderly patients. The disease increased.

After the establishment of the Global Initiative for Asthma (GINA), the diagnosis of asthma is based on a history of variable respiratory symptoms and variable expiratory flow restriction, while the severity of asthma is based on symptoms, history, and lungs. Testing, physical examination and over Minhara. However, asthma is a heterogeneous disease that causes various symptoms and different responses to treatment, and the mechanism for developing asthma is very complicated. Epidemiological studies indicate that these different changes are primarily influenced by race, population, virus and environmental triggers.

Regarding the use of health care, some studies have pointed out that the prevalence of asthma has declined since the drug was administered, but there is still no standard clinical diagnostic criteria for asthma in the world. Traditional methods ignore the severity of asthma in sub-phenotypes and do not fully reflect the comprehensiveness and heterogeneity of the asthma population. In addition, patients are required to go to the clinic regularly to monitor their symptom control, which will impose a huge medical burden on the medical system, patients and their families. It is important to understand the subtype of asthmatic phenotypes.

Numerous studies have used unsupervised cluster analysis to identify new subphenotypes and provide classifications in both adults and children to clarify disease causality and develop new asthma management methods. Cluster analysis is an unsupervised machine learning method that classifies individuals into groups based on the similarity or dissimilarity of variables, increasing homogeneity within clusters or heterogeneity between clusters. Two-stage cluster analysis is a method of combining tiered and non-hierarchical clusters. The first step uses a layered approach to determine the number of clusters and further moves each individual into and between groups through the k-means clustering approach.

In recent decades, the study has conducted cluster analysis in different populations, including severe, difficult to treat, and asthma patients of all severity, attempting to discover and classify subphenotypes and develop new asthma treatments. There is a different pattern of inflammation in the severity of cluster asthma. Moore et al. performed a hierarchical clustering algorithm that selected 15 variables including the percentage of eosinophils and neutrophils to identify 4 clusters and noted that moderate to severe asthma showed neutrophils as Primary or mixed cell inflammatory patterns (Moore, WC, et al., Sputum neutrophil counts are associated with more severe asthma phenotypes using cluster analysis. Journal of Allergy and Clinical Immunology, 2014. 133 (6): p. 1557-1563. E5.). Konno et al. included smokers and found two distinct smokers and current smokers (Konno, S., et al., Distinct Phenotypes of Cigarette Smokers Identified by Cluster Analysis of Patients with Severe Asthma. Annals Of the American Thoracic Society, 2015. 12 (12): p.1771-1780.). However, no studies have used cluster analysis to stratify by gender to explore potentially different asthmatic phenotypes between them.

The present invention explores the effect of gender on the inflammatory response of different mechanisms of the asthmatic phenotype.

The present invention first identifies and characterizes potential phenotypic clusters by sex, and then examines the relationship between these potential asthmatic phenotypes and asthma-related health outcomes in both genders.

The invention includes all lineages of asthma severity, as well as clinical and biological factors measured by sex.

The present invention incorporates 720 asthma patients, including 441 women and 299 men of various asthma severity. Individuals with current or past smoking history are also included. Clinical characteristics, physiological variables and cellular inflammation were assessed and the same eight variables were selected into the unsupervised cluster approach.

The present invention identifies three different clusters with distinctly different characteristics between and between men and women.

The present invention identified three clusters with heterogeneous features in each gender layer and exhibited different asthma phenotypes between males and females. The six clusters identified through clinical and bio-based cluster analysis make sense.

The present invention discloses that different clinical features and clustering patterns of cellular inflammation in six clusters suggest potential information on the potential mechanisms of asthma in terms of gender. The inflammatory pattern is influenced by male smoking status and finally predicts the risk of subsequent health outcomes in the cluster. In women, the allergic tendency of cluster 2 and the obesity and neutrophilia of cluster 3 are characterized by asthma. Significantly high risk of acute attacks; in men, cluster 5 (early asthma, current smokers, low eosinophils) and cluster 6 (long-term asthma duration, smoking cessation, high eosinophils) There are more than three times the risk of asthma worsening in cluster 4.

The present invention discloses a method of classifying a asthmatic phenotype based on gender differences.

The present invention incorporates all levels of asthmatic patients, involving all severity of asthmatic patients.

Taiwan version of the present invention by asthma quality of life assessment questionnaire (Taiwan version of the Asthma Quality of Life, TAQLQ) asthma control and test ^{TM (Asthma Control Test TM, ACT} ) to assess asthma control and quality of life.

The present invention discloses that cluster 1 and cluster 4 have better asthma health outcomes. These two clusters have similar characteristics to older and delayed asthma and exhibit optimal lung function and the lowest percentage of eosinophils and neutrophils. In the poor asthma control group, the female group showed two sub-phenotypic clusters. However, the two subtypes of men are affected by smoking status, unlike female groups. In addition, women can be distinguished by the percentage of neutrophils, but the mean of neutrophils is not significant in male clusters.

The present invention discloses that the concentration of neutrophils may be affected by different body components of women and men. Women have more subcutaneous fat than men, and can increase secretion of two to three times more leptin than abdominal fat. In addition, obese women have higher plasma leptin concentrations than men and cause an increase in neutrophilic inflammation. Accordingly, the present invention discloses that an asthmatic phenotype can be identified by sex.

The present invention covers the smoking population and explores the effects between smoking and inflammatory cells.

The present invention identifies three clusters with heterogeneous features in each gender layer and exhibits different asthma phenotypes between males and females. The six clusters of the present invention are determined to be meaningful through clustering of clinical and biological based cluster analysis.

The present invention discloses the use of discriminate analysis to find eight important predictors that distinguish three types of female asthmatic characteristics, including age, first second forced expiratory volume before bronchodilator use (Pre-FEV ₁ , %) , eosinophils (Log_Eosinophil), immunoglobulin E (Log_IgE), body mass index (BMI), forced vital capacity before bronchodilator use (Pre-FVC, %), neutrophils ( Neutrophil) and Asthma_duration.

The present invention further uses the classification function coefficients (Fisher's linear difference function) to perform the grouping of the female asthmatic characteristics by using the eight important predictors, and inputs the results of the eight examinations of the female patient into the three clustering mode functions for calculation. The highest one is used as the prediction result, and the prediction accuracy of the classification function coefficient is 93.8%.

The present invention uses discriminate analysis to find six important predictors that distinguish three male asthmatic characteristics, including age, first second forced expiratory volume before bronchodilator use (Pre-FEV ₁ ,%), asthma Asthma onset age, forced vital capacity before bronchodilator use (Pre-FVC, %), eosinophils (Log_Eosinophil), body mass index (BMI).

The present invention further uses the classification function coefficients (Fisher's linear difference function) to perform the grouping of the male asthmatic characteristics by using the six important predictors, and the results of the six examinations of the male patient are input into the three clustering mode functions for calculation. The highest one is used as the prediction result, and it is known that the prediction correctness of the coefficient of the classification function is 93.3%.

The invention relates to a grouping method for asthma patients according to different genders, which comprises: A. Establishing a group mode database subsystem for storing a plurality of asthma grouping patterns of different genders, the grouping pattern database The system contains a female asthma patient a group model database and a male asthma patient group pattern database; B. providing a detection subsystem for detecting a plurality of physiological factors of a group of asthma patients, the detection subsystem comprising a gender information collection module, An age information collection module, a first second forced expiratory volume detection module, an eosinophil detection module, a body mass index detection module, and a forced vital capacity detection module; The calculation module is configured to calculate the physiological factors of the group of asthma patients to be divided into groups: c1. input the physiological factors of the group of asthma patients to be divided, and according to the gender of the group of asthma patients, the plurality of asthma in the gender database The patient grouping mode calculates the score of the group of asthma patients to be grouped in each asthmatic group, and c2. selects the highest score among the scores of the grouping asthma patients in each group of asthmatic patients as the grouping result of the group of asthma patients to be grouped; And, D. providing an output module for outputting the grouping result of the patient to be divided into asthma.

In an embodiment, the detection subsystem further comprises an asthma age information collection module, a asthma time course collection module, an immunoglobulin detection module or a neutrophil detection module. group.

In one embodiment, the female asthma patient group pattern database stores a plurality of female asthma patient grouping patterns, wherein the female asthma patient group mode is the age of the female asthma patient, the first second forced expiratory volume, and eosinophils. , immunoglobulin, body mass index, forced vital capacity, neutrophils, and asthma time course are defined.

In one embodiment, the plurality of female asthma patient grouping modes include a grouping mode one, a grouping mode two, and a grouping mode three, wherein the grouping mode has a late onset, a normal BMI, a non-ectopic, and a low neutrophil Characteristics of cells and low eosinophils, group 2 has the characteristics of young, early onset, normal BMI, atopy, high blood eosinophils, low neutrophils, and group 3 has late onset, obesity High neutrophils, low eosinophils and The characteristics of the globulin.

In one embodiment, the grouped result is that the asthmatic patient in cluster mode 2 or group mode 3 is a significantly high risk group of acute asthma attacks.

In one embodiment, the male asthma patient group mode database stores a plurality of male asthma patient grouping patterns, wherein the male asthma patient grouping mode is the age of the male asthma patient, the first second forced expiratory volume, the asthma onset age, Forced vital capacity, eosinophils, and body mass index are defined.

In one embodiment, the plurality of male asthma patient grouping modes include a grouping mode four, a grouping mode five, and a grouping mode six, wherein the grouping mode four has a late-onset, the BMI is normal, the mild feature, and the grouping mode five has a young, The age of onset of asthma is small, current smoking, blood neutrophils are slightly higher, blood eosinophils are low, and grouping mode 6 has late onset, smoking cessation, high blood total immunoglobulin and eosinophils. feature.

In one embodiment, the asthmatic patient whose result is clustered mode 5 or grouped mode 6 is a significantly high risk group of acute asthma attacks.

The invention further relates to a grouping system for asthma patients, comprising: a group mode database subsystem for storing a plurality of asthma patient grouping modes of different genders, the group mode database subsystem comprising a female asthma group group mode data a library and a male asthma patient group pattern database; a detection subsystem for detecting a plurality of physiological factors of a group of asthma patients, the detection subsystem comprising a gender information collection module and an age information collection module a first second forced expiratory volume detecting module, an eosinophil detecting module, a body mass index detecting module and a forced vital capacity detecting module; a computing module for calculating the waiting Grouping the physiological factors of asthmatic patients and then grouping them; and an output module for outputting the to-be-grouped asthma The grouping results of the patients.

In an embodiment, the detection subsystem further comprises an asthma age information collection module, a asthma time course collection module, an immunoglobulin detection module or a neutrophil detection module. group.

In one embodiment, the female asthma patient group pattern database stores a plurality of female asthma patient grouping patterns, wherein the female asthma patient group mode is the age of the female asthma patient, the first second forced expiratory volume, and eosinophils. , immunoglobulin, body mass index, forced vital capacity, neutrophils, and asthma time course are defined.

In one embodiment, the plurality of female asthma patient grouping modes include a grouping mode one, a grouping mode two, and a grouping mode three, wherein the grouping mode has a late onset, a normal BMI, a non-ectopic, and a low neutrophil Characteristics of cells and low eosinophils, group 2 has the characteristics of young, early onset, normal BMI, atopy, high blood eosinophils, low neutrophils, and group 3 has late onset, obesity High neutrophils, low eosinophils and immunoglobulins.

In one embodiment, the grouped result is that the asthmatic patient in cluster mode 2 or group mode 3 is a significantly high risk group of acute asthma attacks.

In one embodiment, the male asthma patient group mode database stores a plurality of male asthma patient grouping patterns, wherein the male asthma patient grouping mode is the age of the male asthma patient, the first second forced expiratory volume, the asthma onset age, Forced vital capacity, eosinophils, and body mass index are defined.

In one embodiment, the plurality of male asthma patient grouping modes include a grouping mode four, a grouping mode five, and a grouping mode six, wherein the grouping mode four has late development and the BMI is normal. Mild characteristics, grouping mode five is young, the age of asthma is small, current smoking, blood neutrophils are slightly higher, blood eosinophils are low, grouping mode six has late onset, smoking cessation, high Characteristics of total blood immunoglobulins and eosinophils.

In one embodiment, the asthmatic patient whose result is clustered mode 5 or grouped mode 6 is a significantly high risk group of acute asthma attacks.

Figure 1. Flowchart for selecting a research object

Figure 2. Blood granulocyte pattern in three female clusters. Subjects were assigned to one of four possible modes depending on the percentage of blood eosinophils (<2% or >2%) and the percentage of sputum neutrophils (<60% or >60%).

Figure 3. Risk ratio (Odds ratio, OR) associated with female asthma related outcomes. A. Acute asthma attack (hospital or emergency visit) B.TAQLQ (cut-off score >61 reflects poor quality of life control) C.ACT (cut-off score <19 reflects poor asthma control). The reference set for logistic regression is cluster 1 (adult onset, mild) based on the lowest TAQLQ and the highest average ACT.

Figure 4. Risk ratio (Odds ratio, OR) associated with male asthma related outcomes. A. Acute asthma attack (hospital or emergency visit) B.TAQLQ (cut-off score >61 reflects poor quality of life control) C.ACT (cut-off score <19 reflects poor asthma control). The reference set for logistic regression regression is cluster 1 (adult onset, mild), which is based on the lowest TAQLQ and the highest average ACT.

The technical solutions in the embodiments of the present invention are clearly and completely described in the following with reference to the accompanying drawings in the embodiments of the present invention. It is obvious that the described embodiments are only one part of the present invention. The embodiments are divided into embodiments, not all. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative efforts are within the scope of the present invention.

Materials and Methods

Research population

Cross-sectional study design attempted to identify the asthma phenotype. All subjects were from July 2012 to August 2015 from the Department of Pulmonary and Critical Care Medicine, Chang Gung Memorial Hospital, Kaohsiung Medical Center, Taiwan. The present invention encompasses subjects over the age of 20 who have been diagnosed with mild to severe asthma by a thoracic physician. The diagnostic criteria were based on the establishment of GINA: (1) a history of respiratory symptoms, (2) an increase in FEV ₁ of more than 12% and 200 ml after inhalation of bronchodilators, and (3) an average daily diurnal PEF variation of >10%, (4) After 4 weeks of anti-inflammatory treatment, FEV ₁ increased by more than 12% and 200 ml from baseline.

After signing the informed consent form, standardized trainers lead the subjects to complete asthma-related questionnaires and lung examinations, and collect their blood samples for demographic information, asthma symptoms, medical history, and biometric data. At the end of the present invention, a total of 720 patients met the criteria for subsequent analysis (Figure 1).

All subjects provided written informed consent before completing the entire process, and the detailed agreement was approved by the Institutional Review Board (IRB) of Kaohsiung Chang Gung Memorial Hospital.

Questionnaire

Asthma related questionnaire

Following the standardized process, after obtaining written informed consent, trained staff interviewed the subjects and completed the questionnaire face to face. The asthma-related questionnaire is divided into five parts, including demographic information, asthma symptoms, health care utilization (HCU), allergic history and medical history, and environmental factors.

The present invention collects demographic information including gender, age, race, education level, and smoking status. Regarding asthma symptoms, there are 4 questions related to the frequency of dyspnea, coughing, sneezing and stuffy nose in the previous year. HCU information comes from the time of hospitalization for asthma or emergency visits. Allergies and medical history provide information about allergic rhinitis, atopic dermatitis and other lung diseases. The final part of the questionnaire was indoor environmental factors, animal skin, cockroaches, mold, burning incense, mosquito repellent, carpet, air conditioning, anti-mite bedding, air purifier and dehumidifier.

Taiwan version of the Asthma Quality of Life (TAQLQ)

The Asthmatic Quality of Life Assessment Questionnaire (AQLQ) is an important component of asthma patients. Changes in patients who responded to the drug or had natural fluctuations (p < 0.001) were measured. AQLQ is an effective evaluation tool with stable measurement characteristics. The 27-point Taiwanese version of the Diaphragm Life Quality Assessment Questionnaire (TAQLQ) is simplified from the original 7-point AQLQ. The simplified AQLQ staff separation reliability of 0.92 (similar to Cronbach's alpha) is considered to provide an acceptable and accurate estimate for asthmatic patients, especially those with less education in southern Taiwan.

The patient is provided with TAQLQ and his score is calculated. The higher the score, the worse the quality of life of asthma patients. There are five levels (61, 89, 106, and 120) based on the cut-off score, which means fairly mild, mild, moderate, or severe asthma. Prior to statistical analysis, the total score will be converted to a binary variable based on score 61, which represents good control and poor control of quality of life, respectively.

Asthma Control Test ^{TM (Asthma Control Test TM, ACT} )

The ACT is a patient-based 5-point questionnaire that assesses asthma control over the past four weeks. GINA defines that a cut-off score of <19 reflects poor asthma control. The ACT assessment is used to assess routine clinical visits or poor control in research due to its ease of guidance. The score of the asthma patient.

ACT has high reliability and effectiveness, its internal consistency reliability is 0.84, and it is found that the expert rating has the highest significant correlation coefficient with ACT score (r=0.45, p=.0001). Similar findings were found in longitudinal studies and cross-sectional studies. In the present invention, the score is converted to a binary variable (<19 or 19), representing poorly controlled asthma and well-controlled asthma groups, respectively.

Inflammatory pattern

Based on the percentage of eosinophils and neutrophils greater than or less than the median, sputum was identified to have four granulocyte patterns. The definition of an inflammatory pattern provides more meaningful information about these granulocytes. In the present invention, although the granulocytes in the blood are measured, the same method according to the median of eosinophils and neutrophils is used. In the female cluster, the median percentage of eosinophils and neutrophils was 2% and 40%, respectively. In the male population, the median percentage of eosinophils and neutrophils was 2.4% and 40.4%, respectively. The four inflammatory patterns are defined as follows in Table 1 (female):

Research measurement

Anthropometric examination

Physiological tests include height, weight, body mass index (BMI), waist circumference, hip circumference and waist-to-hip ratio (WHR). On the same day of clinical visit, the subject's body composition was measured by using TANITA (BF-720) FAT MONITOR. BMI is calculated by standard algorithms: weight (kg) / height (m2). Waist and hip circumference are obtained by the staff with a measuring tape. The value of WHR is calculated from the waist/hip circumference.

Pulmonary function

All lung examination data were obtained from the hospital medical record system. The professional respiratory therapist follows a standard protocol to measure the “baseline” spirometry before and after bronchodilator use, and to assess the highest value of Forced vital capacity (FVC), forced expiratory volume in the first second (Forced expiratory volume) In one second, FEV ₁ ) and FEV ₁ /FVC ratio at least three times.

Blood sample

Blood samples were collected on the day of the subject's clinical visit and analyzed by hospital laboratory staff. Assessment of atopic status by measuring total serum immunoglobulin (Total serum IgE) and inflammatory cells, including white blood count (WBC), eosinophils, neutrophils, monocytes, lymph Percentage of cells and basophils. All data comes from the hospital's medical record system.

Variables and definitions

Initially, there were 273 variables in the asthma-related questionnaire. After deleting the text form problem, dealing with redundant variables and exclusion variables, and the deletion rate was greater than 10%, the 26 variables were further statistically analyzed. The variable process and definition are as follows: Childhood asthma is defined as age at onset <16 years, and obesity is defined as BMI 30 kg / m ^ 2 . Allergic rhinitis, a sexual dermatitis, asthma symptoms, are binary variables. Among them, the subjects reported that they were diagnosed as rhinitis and atopic dermatitis by the doctor as "Yes", and if not, the code was "No". BMI, WHR, body fat, age of asthma attack, duration of asthma, pre-FEV ₁ % predictive value before bronchodilator use, pre-FVC% predictive value before bronchodilator use, FEV ₁ /FVC ratio, total immunization Globulin, neutrophils, lymphocytes, monocytes, eosinophils, basophils, and WBC are continuous variables. The definition of high-dose inhaled corticosteroids (ICS) is equivalent to budesonide > 800 μg/day and fluticasone > 500 μg/day. The asthma medication controller includes ICS, OCS, theophylline and leukotrienes.

The asthma health outcome variables were assessed as follows: (1) Acute asthma attacks, including hospitalization and emergency visits in the past year. (2) TAQLQ scored within 2 weeks of asthma quality of life score, with 61 as the decision score to define whether the quality of life was good or poor, (3) ACT was determined by the subject self-reported asthma control for 4 weeks, and 19 as a decision score is defined as whether the asthma control is good or poor.

The last eight variables are the selection variables in the two-stage cluster analysis. TAQLQ, ACT and acute exacerbation of asthma (hospital hospital for asthma and emergency visits) were selected as outcome variables.

Statistical Analysis

In order to examine the information on the distribution of variables, it was observed that some variables exhibited a very general distribution, such as total serum immunoglobulin, percentage of blood eosinophil count, and percentage of basophil count. To improve the normality of the distribution, these variables were log transformed for subsequent statistical analysis. Regarding the missing data processing, the missing rate of the variable is excluded from the study if it is greater than 10%, and is estimated by the multiple interpolation method if the missing rate is 10% or lower.

Perform orthogonal maximum factor analysis to reduce and summarize many variables to provide variable selection for cluster analysis. Extract the common factor between variables and replace the original complex multivariate structure with a smaller dimension (factor). A total of 12 and 13 variables (female and male) consisted of 5 factors, and one or two representative variables were selected from each factor for cluster analysis. Eight variables included (1) age, (2) BMI, (3) age of onset of asthma, (4) pre-FEV ₁ % predicted before bronchodilator use, and (5) pre-FVC before bronchodilator use % predictive value, (6) total serum immunoglobulin, (7) percentage of eosinophils in the blood, and (8) percentage of blood neutrophils further provide a two-stage cluster analysis.

A two-stage cluster analysis was performed to classify subjects into larger groups based on similarity or dissimilarity of variables. Stepwise discriminant analysis to determine the most influential cluster parameters between the three clusters. In order to find differences between male and female subjects in the three clusters, X ² number of data ANOVA (One-way ANOVA tests) for continuous variables and categorical variables, respectively and used. The risk of asthma-related outcomes in the cluster was assessed by using multivariate logistic regression. Data was analyzed using SPSS version 20.0 and p values less than 0.05 were considered significant.

result

Demographics of the study subjects

In the initial data set, 747 subjects diagnosed with asthma were recruited from Chang Gung Memorial Hospital in southern Taiwan. A total of 421 subjects (96.3%) were eligible for 421 women and 299 male adults with asthma according to exclusion criteria, 58% of whom were women.

Most of the subjects in the present invention are asthma with adult onset. The percentage of men with early-onset asthma (10%) is higher than that of women. The percentage of women with obesity (BMI > 30 kg / m ^ 2, 15%) was higher than that of men, compared with the higher male mean of WHR. About half (45.3%) of male subjects have a history of smoking. Compared with women, the average total concentration of serum immunoglobulin in men is slightly higher. The lung function of women before bronchiectasis is slightly higher than that of men. The proportion of comorbidities in female and male subjects is similar. The incidence of asthma symptoms in the past year was higher in women than in men. Most (44.7%) female subjects used low to moderate doses of ICS, while male subjects used higher percentages of high dose ICS (44.8%).

factor analysis

Orthogonal maximal factor analysis identified five factors in women and men (Table 2). Select one or two representative variables from each dimension (factor) for cluster analysis. Factor analysis has similar results in both women and men.

The final 8 variables were chosen: age, BMI, age of onset of asthma, pre-FEV ₁ % predicted using bronchodilator (Pre-FEV ₁ ), pre-FVC% predictive value (Pre-FVC), total serum immunization Globulin, blood eosinophil percentage and blood neutrophil percentage were clustered. Among the two genders, the three most influential factors are body composition parameters (BMI, waist circumference and body fat), some blood biomarkers (lymphocytes and neutrophils) and lung function parameters (Pre-FEV). ₁ predicted value and Pre-FVC predicted value) feature.

Stepwise discriminant analysis

A stepwise discriminant analysis using the same 26 variables presented 9 and 8 of the strongest discriminating variables assigned to the cluster in women and men. Among the two genders, age and pre-FEV ₁ % predicted before using bronchodilators were the first and second most influential variables (F=209.3, P<0.0001 and F=237.1, p<0.0001; F= 192.5, P < 0.0001 and F = 167.6, p < 0.0001).

In the female cluster, the nine discriminant variables are age, pre-FEV ₁ % predicted before bronchodilator use, blood eosinophils, total immunoglobulin, BMI, FEV ₁ /FVC ratio, blood neutrophils Cells, asthma onset age and blood basophils. In the male cluster, the six discriminant variables were age, pre-FEV ₁ % predicted before bronchodilator use, age at onset of asthma, predicted FVC% before bronchodilator use, blood eosinophils and BMI.

Cluster analysis

Three clusters were identified among women and men through a two-stage cluster analysis of the subjects.

Reference cluster

In the female and male clusters, Cluster 1 and Cluster 4 present the best cluster of asthma control with the highest lung function and subsequent TAQLQ in, for example, acute asthma exacerbations Best health results on the ACT and on the ACT. Patients with a higher ACT score have higher lung function. These two groups represent relatively mild asthma symptoms, with clinical features of the elderly, normal BMI, and asthma in adults. Both cluster 1 and cluster 4 have the lowest total immunoglobulin and the lowest percentage of eosinophils neutrophils and are composed primarily of oligoparticle globular patterns. Oligoparticle globularity is the most common pattern observed in mild-to-moderate asthma clusters.

Female cluster characteristics

The three clusters showed significant differences in clinical features and inflammatory patterns. Cluster 1 and Cluster 2 have significant differences in clinical features and inflammatory patterns, indicating that significant heterogeneity between the two clusters can be observed. Conversely, Cluster 1 and Cluster 3 appear to have some similar clinical features, such as age and adult onset asthma, whereas in BMI, body fat, waist circumference, and WHR physical examination can be divided into two groups. Cluster 3 was defined as an obese group with minimal lung function and poor asthma control.

Cluster 1 (N=174). Cluster characteristics: late onset, normal BMI, non-atopic, low neutrophils, low eosinophils

The characteristics of cluster 1 are taken as the grouping mode 1 of the group of asthma patients. This cluster was the largest cohort (N=174) and consisted of slightly older subjects (mean 57.95 years) with approximately normal subjects with a mean BMI (mean, 25.1 kg/m2) , body fat and delayed onset of WHR (age age > 16 years). The lung function of cluster 1 had the highest average of FEV ₁ % predicted values (average, 99.2%) and FEV ₁ /FVC ratio (average, 79%). Compared with other clusters, this group of serum total immunoglobulin (80.79 IU / ml), blood eosinophil count (mean, 2.34%) and blood neutrophil count (average 58.77%) in three The lowest among the clusters. There are different inflammatory cell patterns between the three female clusters. This group has a lower incidence of asthma symptoms in the past year. In terms of drugs, 33.3% of the subjects had high-dose inhaled corticosteroids, 12.7% of the subjects had OCS, and the first group had fewer than the second and third groups. Subjects in Group 1 had the lowest TAQLQ score and the highest ACT score. Compared to cluster 2 and cluster 3, cluster 1 is a relatively good asthma control group.

Cluster 2 (N=117). Cluster characteristics: young, early onset, normal BMI, atopy, high blood eosinophils, low neutrophils

The characteristics of cluster 2 are taken as the grouping mode 2 of the group of asthma patients. The mean age of the subjects in this cluster (37.56 years) and the age of onset of asthma (30.28 years) were the smallest. 10.3% of subjects were early-onset asthma (age of onset) 16 years old). Cluster 2 subjects had the worst BMI, body fat, waist circumference and WHR mean. The average percentage of pre-FEV ₁ (average, 87.3%) and FEV ₁ /FVC ratio (average, 78.9%) before bronchodilator use was normal. The eosinophil count was highest in the atopic asthma group (mean 3.70%) and the total serum immunoglobulin level (average 312.55 IU/ml) was highest in the three groups. Half of these subjects (53.0%) were self-reported allergic rhinitis. It is worth noting that this group of blood neutrophil counts (average 59.8%) is lower. The incidence of asthma symptoms was slightly higher in this group, and 63.8% of the subjects reported usual sneezing symptoms. The lowest percentage of subjects (24.8%) received high-dose ICS, however, 21.4% of subjects received OCS, which was the most common of the three groups. More than half (59%) of the subjects used antihistamines.

Cluster 3 (N=130). Cluster characteristics: late onset, obesity, high neutrophils, low eosinophils and immunoglobulins

The characteristics of cluster 3 are taken as the grouping mode 3 of the group of asthma patients. Compared with cluster 2, cluster 3 is asthmatic in the elderly and adults. This group with the highest BMI and body fat mean was defined as the obese group. In cellular inflammation, this group has the highest percentage of blood neutrophils, but has the lowest percentage of total immunoglobulin and eosinophils. Cluster 3 is characterized by non-eosinophilic asthma. Cluster 3 has a higher risk of acute asthma attacks, poor asthma control, and worsening lung function. Display consisting mainly of neutrophils and mixed granulocyte patterns The subgroup presented with poor lung function and more drug use. Since adipose tissue produces proinflammatory mediators such as leptin, TNFα and IL-6, leptin stimulates the release of TNFα by monocytes, which activates neutrophils and can aggregate neutrophils into the airways. Activated neutrophils release mediators such as neutrophil elastase and cause contraction of human airway smooth muscle (hASM), resulting in narrowing of the airways. In addition to receiving high doses of corticosteroids, NEA patients still respond poorly to corticosteroids. This means that Cluster 3 uses a higher proportion of ICS and is less healthy in asthma.

Cluster 1 and Cluster 3 in the female cluster have different inflammatory patterns (Fig. 2), indicating that obesity not only affects subsequent health outcomes, but also has different biological pathways. In multivariate logistic regression, cluster 2 and cluster 3 are the worst asthma control groups compared to cluster 1. There is significant heterogeneity in both clinical features and inflammatory patterns in both clusters. Two biological pathways for asthma can be seen in Cluster 2 and Cluster 3.

Male cluster characteristics

Cluster 4 (N=112). Cluster characteristics: late, BMI normal, mild

The characteristics of cluster 4 are taken as the grouping mode 4 of the group of asthmatic patients. The subjects in this cluster were older (mean 62.45 years) with a normal BMI, body fat, waist circumference and waist-to-hip ratio. All subjects were delayed onset (age age > 16 years). This group had a short duration of asthma (average 2.89 years), unlike cluster 5 and cluster 6. The percentage of subjects with a history of smoking in this group was small, and only 9.9% of the subjects were current smokers. Cluster 4 had the highest pre-FEV ₁ mean (average, 94.59%) and FEV/FVC ratio (mean, 76%) before bronchodilator use compared to other male clusters. The group had the lowest serum total immunoglobulin (136.19 IU/ml) and blood eosinophil count (mean 2.92%), and the two biomarkers were significantly lower than cluster 6. Although the frequency of asthma symptoms did not differ significantly in the male population, this group of subjects appeared to have less asthma symptoms during sleep (28.6%). 9.8% of subjects used oral corticosteroids at a lower rate than cluster 5. This group has the lowest TAQLQ score and the highest ACT score. Like cluster 1 in women, this group was defined as a relatively good asthma control group.

Cluster 5 (N=92). Cluster characteristics: young, age of asthma, small smoking, current smoking, blood neutrophils slightly higher, blood eosinophils are low.

The characteristics of cluster 5 are taken as the grouping mode 5 of the group of asthma patients. This age group (mean 39.06 years old), the group with the lowest age of onset of asthma (mean 29.30 years) has a long duration of asthma. Cluster 5 has a slightly higher BMI average. 22.4% of subjects were obese (BMI 30 kg / m ^ 2 ), but the body fat content between male clusters is not significant. Compared with other male clusters, the cluster has the highest percentage of current smokers (28.4%). The average percentage of pre-FEV ₁ before bronchodilator use in lung function was 86.22%, and the FEV ₁ /FVC ratio (average, 78%) was normal in this group. 40.5% of the subjects were self-reported allergic rhinitis. The group showed high serum total immunoglobulin levels (277.83 IU/ml) and blood neutrophils (62.14%), but the percentage of blood eosinophils was higher. low. The inflammatory pattern is different from that of the female group, which is defined as a young and ectopic state. Although there was no significant difference in asthma symptoms between the three groups, the subject had a higher percentage of daily asthma symptoms. More than half of the subjects (62.1%) received ICS and 23.5% required OCS. In general, Group 5 was defined as young, current smokers and low blood eosinophil counts. In the male cluster, cluster 5 and cluster 6 showed a higher risk of acute asthma attacks compared to cluster 4 (reference group), which had similar self-reported asthma control. Subjects in cluster 5 and cluster 6 had significant differences in clinical characteristics and blood eosinophil percentage. 28.4% of the cluster 5 individuals who were young and had an early onset of asthma were current smokers. In the inflammatory cell mode, this group has a slightly higher percentage of blood neutrophils and a lower percentage of blood eosinophils. However, cluster 6 with the oldest age and adult onset of asthma had the highest percentage (56.3%) of the smoking population, including smokers and smokers. The majority (43.7%) of the individuals were smokers. Subjects in this group had low FEV ₁ /FVC (mean, 0.69%) and the highest percentage of blood eosinophils, meaning that significant heterogeneity of the two smoking clusters could be observed. This cluster indicates the presence of two different phenotypes in current smokers and smokers who are severely asthmatic.

28.4% of the subjects were current smokers, which is the highest proportion of the three clusters. Although subjects in cluster 5 had a higher percentage of blood neutrophils, this was not statistically different among the three male clusters. However, according to the inflammatory cell pattern, this group had the highest percentage of neutrophil-based patterns in the three male clusters. It is shown that smoking is associated with neutrophil inflammation, but does not aggravate eosinophilic inflammation. Since cigarette smoke partially induces Th-17 type inflammation, Th-17 cytokines recruit neutrophils in the airways and activate IL-8 from epithelial cells, thus supporting the characteristics of cluster 5. The proportion of blood eosinophils in cluster 5 is reduced compared to cluster 6 which is defined as a non-smoking group. In addition, the risk of asthma in cluster 5 was 2.23, and the proportion of OCS used was the highest. Compared with smokers and non-smokers, patients with severe asthma who smoked had a poorer prognosis.

Cluster 6 (N=77) late-onset, quitters, high blood total immunoglobulins and eosinophils

The characteristics of cluster 6 are taken as the grouping mode of grouping of asthma patients. This cluster consisted of older subjects (mean 65.56 years) and long duration asthma (mean 10.37 years). Similar to cluster 4, most subjects had delayed asthma (mean 55.20 years). The BMI, body fat, waist circumference and WHR were normal in this group, only one subject was obese (BMI 30 kg / m ^ 2 ). This group had the highest percentage of smoking history compared to cluster 4 and cluster 5 and showed the highest percentage (43.7%) of quitters. Subjects in cluster 6 had a decrease in lung function with pre-FEV ₁ % predicted values (mean, 65.43%) and FEV/FVC (mean, 69%) before bronchodilator use. In the inflammatory mode, cluster 6 had the highest blood eosinophils and the total serum immunoglobulin level (252.21 IU/ml) was moderately elevated. These subjects reported similarities to cluster 4 and received 9.9% more corticosteroid treatment than two asthma medical controllers. Cluster 6 consists of elderly patients with a history of smoking and eosinophilic airway inflammation. 43.7% of the subjects were smokers. It was found that the quitters had an eosinophil phenotype. The current mechanism of eosinophilic inflammation in smokers and smokers is mainly through the induction of smoking-induced Thymic stromal lymphopoietin (TSLP) and then promotes the T _H 2 immune response. Regarding lung function, cluster 6 has poor lung function (FEV ₁ /FVE ratio, mean 69%) may be due to older age from this group of subjects, due to impaired lung function in the elderly, decreased chest wall compliance and supporting tissue Such as the reduction of elastic fibers, caused by damage around the alveolar duct. Smokers are predominantly non-eosinophilic sputum phenotypes, while quitters have eosinophil phenotypes similar to cluster 5 and cluster 6. Changes in the pattern of inflammation in the blood indicate that smoking conditions may affect the inflammatory phenotype of male patients.

Cluster and asthma control related results

In the multi-logist regression analysis, acute asthma attacks, TAQLQ, and ACT were selected as asthma-related outcomes. An acute asthma attack is defined as a hospitalization or emergency visit last year. TAQLQ and ACT are converted to binary variables according to decision scores 61 and 19, respectively. In the female and male populations, cluster 1 and cluster 4 were selected as reference groups based on the lowest frequency of acute asthma attacks (hospital or emergency department visits), the lowest score of TAQLQ and the highest score of ACT.

In the female cluster, the risk of acute asthma attacks in cluster 2 (early asthma, atopic) and cluster 3 (late asthma, obesity) was significantly more than 2-fold (OR=2.51, 95% CI=1.12~5.59 and OR = 2.22, 95% CI = 1.01 to 4.93, Figure 3A). Patients with cluster 2 (early asthma, atopic) showed poor quality of asthma compared with cluster 1 (TAQLQ, OR=1.69, 95% CI=1.01 to 2.89, Figure 3B), but these were poorly controlled. There was no significant difference in asthma (ACT, OR) = 1.01, 95% CI = 0.60 to 1.73, Figure 3C). Cluster 3 (delayed asthma, obesity) was associated with an increased risk of asthma control (ACT, OR = 1.74, 95% CI = 1.01 to 3.00, Figure 3B), but there was no significant association between asthmatic quality of life (TAQLQ) (figure 3C).

In the male cluster, cluster 5 (current smokers, low eosinophils) and cluster 6 (smokers, eosinophils) have a marginal trend that increases more than three times the risk of acute asthma attacks (OR= 3.12, 95% CI = 0.97 to 9.97 and OR = 3.43, 95% CI = 0.99 to 11.84, Figure 4A). Cluster 5 (current smokers, low eosinophils) had a significantly increased risk of poor quality of asthma (TAQLQ, OR=2.23, 95% CI=1.10-4.50, Figure 4B), an increased risk of asthma control >50% ( ACT), OR = 1.56, 95% CI = 0.84 to 2.89, Figure 4C), but the results of ACT did not reach statistical significance. There were no significant differences between cluster 6 (smokers, eosinophils) and reference groups in TAQLQ and ACT (TAQLQ, OR=1.42, 95% CI=0.62 to 3.29, Figure 4B, and ACT, OR=1.005) , 95% CI = 0.48 to 2.12, Figure 4C)

Group mode establishment

Eight important predictors for distinguishing three types of female asthmatic characteristics using discriminate analysis include age, first-second forced expulsion before bronchodilator use (Pre-FEV ₁ ,%), eosinophils Cells (Log_Eosinophil), immunoglobulin E (Log_IgE), body mass index (BMI), forced vital capacity before bronchodilator use (Pre-FVC, %), neutrophil (Neutrophil), and asthma course (Asthma_duration), as shown in Table 3.

The eight important predictors were further used to classify the characteristics of female asthma patients by the classification function coefficient (Fisher's linear difference function). The results of the eight examinations of female patients were input into the three cluster mode functions, and the highest score was used. Predict the result and know that the prediction correctness of this classification function coefficient is 93.8%.

Six important predictors for distinguishing three male asthmatic features using discriminate analysis include age, first-second forced expiratory volume before bronchodilator use (Pre-FEV ₁ ,%), age of onset of asthma (Asthma onset age), forced vital capacity before pre-bronchodilator (Pre-FVC, %), eosinophils (Log_Eosinophil), body mass index (BMI), as shown in Table 5.

The six important predictors were further used to classify the characteristics of male asthma patients by the classification function coefficient (Fisher's linear difference function). The results of the six tests of male patients were input into three cluster mode functions, and the highest score was used. Predict the result and know that the prediction correctness of this classification function coefficient is 93.3%.

Embodiment 1

The age of an a woman with asthmatic female A (56 years old), the first second forced ventilatory volume before bronchodilator (Pre-FEV ₁ ,%), eosinophils (Log_Eosinophil), immunoglobulin E (Log_IgE), Body mass index (BMI), forced vital capacity (Pre-FVC, %), neutrophil (Neutrophil), and Asthma_duration before bronchodilator use are as follows (Table 7):

Substituting Ms. A's data into the grouping mode can be obtained: -96.332+0.712*age+1.246*BMI+2.185*Log_IgE+5.652*Log_Eosin+0.080*pre_FEV_p+0.574*pre_FVC_p+0.791*Neutrophil-0.200*Asthma_duration

Substituting Ms. A's data into grouping mode 2 can be obtained: -81.019+.451* age+1.125*BMI+3.857*Log_IgE+8.401* Log_Eosin+.060*pre_FEV_p+0.544*pre_FVC_p+0.802*Neutrophil-0.140*Asthma_duration

Substituting Ms. A into subgroup mode 3: -91.828+0.788*age+1.372*BMI+3.085*Log_IgE+5.896*Log_Eosin-0.017*pre_FEV_p+0.458*pre_FVC_p+0.871*Neutrophil-0.176*Asthma_duration

After calculation, the grouping mode one is 97.57 points, the grouping mode two is 99.29 points, and the grouping mode three is 105.40 points. Since the grouping mode three has a prediction score of 105.40, it is the highest, so Ms. A is judged to belong to the cluster 3, which is female. Asthma has the characteristics of obesity and high-characteristic eosinophils, and the correctness of this cluster 3 prediction is 99.2%.

Embodiment 2

The age of Mr. B (30 years old) of a male patient, the first second forced exhalation before bronchodilator (Pre-FEV ₁ ,%), the age of onset of asthma (Asthma onset age = 30 years), use of bronchodilator The previous data on forced vital capacity (Pre-FVC, %), eosinophils (Log_Eosinophil), and body mass index (BMI) are as follows (Table 8):

Substituting Mr. B's data into the grouping mode 4 can be obtained: -98.123+0.644*age+0.109* asthma onset age+2.220*BMI+3.753*Log_Eosinophil+0.109*pre_FEV_p+0.802*pre_FVC_p

Substituting Mr. B's data into the grouping mode can be obtained: -79.718+0.446*age-0.036* asthma onset age+2.326*BMI+2.209*Log_Eosinophil+0.159*pre_FEV_p+0.712*pre_FVC_p

Substituting Mr. B's data into the grouping mode can be obtained: -79.669+0.715*age+0.090* asthma onset age+2.078*BMI+5.390*Log_Eosinophil+.004*pre_FEV_p+.679*pre_FVC_p

After calculation, the grouping mode 4 is 66.18 points, the grouping mode 5 is 70.65 points, and the grouping mode 6 is 63.23 points. Since the grouping mode is divided into the highest level of 70.65, Mr. B is judged to belong to the cluster 5, which is pumped for male asthma. The characteristics of this smoker and low eosinophils, this cluster 5 prediction accuracy is 94.0%.

A person skilled in the art can understand that all or part of the process of implementing the above embodiments can be completed by a computer program to instruct related hardware, and the program can be stored in a computer readable storage medium. The program, when executed, may include the flow of an embodiment of the methods as described above. The storage medium may be a magnetic disk, a optical disk, a read-only memory (ROM), or a random access memory (RAM).

The above is a preferred embodiment of the present invention, and it should be noted that those skilled in the art can also make several improvements and retouchings without departing from the principles of the present invention. It is the scope of protection of the present invention.

Claims (8)

A method for grouping asthma patients according to different genders, comprising: A. Establishing a group mode database subsystem for storing a plurality of asthma grouping patterns of different genders, the group mode database subsystem comprising a female A grouping model database of asthma patients and a group model database of male asthma patients; B. providing a detection subsystem for detecting a plurality of physiological factors of a group of asthma patients, the detection subsystem comprising a gender information collection model a group, an age information collection module, a first second forced expiratory volume detection module, an eosinophil detection module, a body mass index detection module, and a forced vital capacity detection module; Providing a calculation module for calculating the physiological factors of the group of asthma patients to be divided into groups: c1. inputting the physiological factors of the group of asthma patients to be divided, according to the gender of the group of asthma patients to be divided into the gender database The grouping pattern of asthma patients calculates the score of the group of asthma patients to be grouped in each asthmatic group, and c2. Select the group to be treated with asthma Clustering in the score of each mode asthma patients to be the highest score as the result of clustering asthma patients of clustering; and, D providing an output module for outputting the asthmatic patients to be clustering result of clustering. The method for grouping asthma patients according to different genders according to claim 1, wherein the detection subsystem further comprises an asthma age information collection module, a asthma time course collection module, and an immunoglobulin detection Test module or a neutrophil detection module. The method for grouping asthma patients according to different genders according to claim 2, wherein the female asthma group group mode database stores a plurality of female asthma grouping patterns, wherein The female asthmatic grouping pattern is defined by the age of female asthma patients, forced expiratory volume in the first second, eosinophils, immunoglobulins, body mass index, forced vital capacity, neutrophils, and duration of asthma. The method for grouping asthma patients according to different genders according to claim 3, wherein the plurality of female asthma patients group mode includes a group mode 1, a group mode 2, and a group mode 3, wherein the group mode 1 has late-onset, Normal BMI, non-atopic, low neutrophils, low eosinophils, grouping pattern 2 with young, early onset, normal BMI, atopy, high blood eosinophils, low neutrophil The characteristics of granulocytes, grouping mode 3 are characterized by late onset, obesity, high neutrophils, low eosinophils and immunoglobulins. The asthma patient grouping method according to claim 4, which is grouped according to different genders, wherein the grouping result is a significant high-risk group of acute asthma attacks in the group mode 2 or the group mode 3. The grouping method for asthma patients according to claim 2, wherein the male asthma group group mode database stores a plurality of male asthma grouping patterns, wherein the male asthma group is modeled by male asthma patients. Age, first second forced expiratory volume, asthma onset age, forced vital capacity, eosinophils, and body mass index are defined. The method for grouping asthma patients according to different genders according to claim 6, wherein the plurality of male asthma patients group mode comprises a group mode 4, a group mode 5, and a group mode 6, wherein the group mode 4 has late development, BMI Normal, mild characteristics, grouping mode 5 is young, the age of asthma is small, current smoking, blood neutrophils are slightly higher, blood eosinophils are low, and grouping mode 6 has late-onset, smoking cessation High blood total immunity Characteristics of globulin and eosinophils. The asthma patient grouping method according to claim 7 according to claim 7, wherein the grouping result is a significant high-risk group of acute asthma attacks in the group mode 5 or the group mode 6.