Construction method of model for predicting risk of complicated oral mucositis during hematopoietic stem cell transplantation of leukemia patient

Abstract

The invention discloses a construction method of a leukemia patient hematopoietic stem cell transplantation concurrent oral mucositis risk prediction model, which comprises the following steps: estimating a sample size by adopting an EPV principle, and collecting enough acute leukemia patients for hematopoietic stem cell transplantation treatment cases, basic information, clinical symptoms, hematology indexes and outcome indexes including cases are collected, data arrangement and statistical analysis are carried out by using SAS9.4, risk factors are obtained, and Plt; and the risk factor variable of 0.05 is incorporated into a multi-factor Cox regression model, a risk prediction model is constructed, and visualization is realized. Through verification, the model is stable. The data source of the model has the characteristics of being easy to obtain, objective in index and small in influence on patients, and the model is more suitable for clinical first-line medical staff to monitor oral mucositis.

Description

Technical Field

[0001] The present invention belongs to the field of clinical medicine technology, and specifically relates to a method for constructing a risk prediction model for oral mucositis during hematopoietic stem cell transplantation in leukemia patients, especially acute leukemia patients. Background Art

[0002] Hematological tumors mainly include three major categories of diseases, including leukemia, lymphoma and multiple myeloma. Among them, leukemia is an important disease among malignant tumors of the hematological system. According to data from the China Cancer Center, the incidence of malignant tumor leukemia in China in 2022 is 4.54 / 100,000, ranking 13th in the incidence of malignant tumors and 12th in mortality. It is the primary factor leading to death from malignant tumors in young people under 35 years old. Leukemia can be divided into acute leukemia and chronic leukemia according to the onset and the stage of tumor cells. Among them, acute leukemia is a highly invasive malignant tumor derived from the hematopoietic system, mainly including acute myeloid leukemia and acute lymphocytic leukemia. In recent years, although new therapies such as targeted therapy and immunotherapy for patients with hematological tumors have been continuously developed, except for acute promyelocytic leukemia, hematopoietic stem cell transplantation is still the core strategy for treating high-risk groups or relapsed / refractory acute leukemia patients. After hematopoietic stem cell transplantation, patients with acute leukemia can re-establish normal hematopoietic and immune functions. Oral mucositis is a common complication of hematopoietic stem cell transplantation.

[0003] After pretreatment, hematopoietic stem cell transplant patients are prone to complications such as bleeding, anemia, infection, and organ damage. Among them, infection is the most common complication, and oral mucositis, as an important component of infectious complications, is one of the most common complications. The incidence of oral mucositis in hematopoietic stem cell transplant patients is as high as 70%, which is a predisposing factor for a variety of adverse prognoses, so oral mucositis needs to be taken seriously.

[0004] The clinical manifestations of oral mucositis include pain, foreign body sensation in the mouth, difficulty opening the mouth, and the formation of white spots. After oral mucositis occurs, the patient's physiological functions such as language communication, oral eating or drinking are seriously affected due to pain, resulting in a decrease in the patient's quality of life, which in turn affects the intake of nutrients, changes in comfort, and delays recovery time. Oral mucositis not only affects the patient's quality of life, but may also lead to other serious clinical consequences. The destruction of the oral mucosal barrier increases the risk of oral bacterial infection. Oral colonizing bacteria can easily enter the blood vessels and cause bloodstream infection, which may cause systemic infection or induce graft-versus-host disease, indirectly increasing the mortality rate of hematopoietic stem cell transplant patients, prolonging the patient's hospitalization time, and increasing the economic burden on patients.

[0005] Research Progress on Influencing Factors of Oral Mucositis. At present, there is still a lack of accurate models for predicting oral mucositis clinically, and most rely on traditional risk factor analysis or scales for evaluation. Existing studies have shown that the occurrence of oral mucositis is affected by multiple factors, such as: (1) Patient characteristics: age, gender, underlying diseases, oral hygiene status; (2) Treatment-related factors: pretreatment regimens (chemotherapy and radiotherapy doses), type of transplantation (autologous or allogeneic), application of immunosuppressants;

[0006] (3) Genetic and biological factors: levels of cytokines (such as IL-6, TNF-α), antioxidant capacity, gene polymorphisms;

[0007] (4) Hematological indexes: results of blood routine tests, liver function tests, electrolyte tests.

[0008] There are currently three different oral assessment scales, and the comparison is as follows:

[0009] One is the assessment scale developed by the World Health Organization (WHO). This scale divides the severity of oral mucositis into 4 grades. Researchers believe that the severity of mucositis has little correlation with the size of the ulcer surface, but is based on the assessment of the patient's food intake ability. The drawback of this scale is that the severity of mucositis has little correlation with the size of the ulcer surface, but is based on the assessment of the patient's food intake ability.

[0010] The second is the Oral Assessment Guide (OAG). This scale divides the severity into 3 grades, and there are 8 assessment items. Among them, the subjective items are voice and swallowing ability; the objective aspects include the assessment of lips, tongue, saliva, mucosa, gums and teeth / dentures. The drawback is that there are many items, and the severity of oral mucositis may be different for different patients with the same score, and the pain felt subjectively by the patient is not evaluated.

[0011] The third is the Oral Mucositis Assessment Scale (OMAS). This scale divides the oral cavity into 9 regions, can effectively track the change process of mucositis over time, and also evaluates the patient's dysphagia and pain. The drawback is that this scale has many items and is not applicable to patients who cannot open their mouths for oral examination due to pain after hematopoietic stem cell transplantation.

[0012] A risk prediction model is a tool constructed by combining a large amount of clinical data and biomarkers and using statistical methods, aiming to predict the risk of disease occurrence, progression, or complications. In patients with acute leukemia, especially during hematopoietic stem cell transplantation, constructing an accurate risk prediction model for oral mucositis is of great significance. The prediction model not only helps to identify high-risk patients, but also provides a scientific basis for early intervention in acute leukemia patients undergoing hematopoietic stem cell transplantation, and is of important value in reducing the incidence of complications, improving poor prognosis, and enhancing the quality of life of patients. Summary of the Invention

[0013] The object of the present invention is to provide a method for constructing a risk prediction model for oral mucositis during hematopoietic stem cell transplantation in patients with acute leukemia.

[0014] The following implementation schemes are provided to achieve the object of the present invention.

[0015] In one implementation scheme, a method for constructing a risk prediction model for oral mucositis during hematopoietic stem cell transplantation in patients with acute leukemia of the present invention includes the following steps:

[0016] 1) Estimate the sample size using the empirical principle of sample size calculation (EPV) for the model development cohort;

[0017] 2) Collect a sufficient number of case samples of acute leukemia patients undergoing hematopoietic stem cell transplantation and divide them into an oral mucositis group and a non-oral mucositis group;

[0018] 3) Collect the basic information, clinical symptoms, hematological indicators, and outcome indicators of acute leukemia patients undergoing hematopoietic stem cell transplantation;

[0019] 4) Use SAS 9.4 for data collation and statistical analysis, and obtain risk factors through univariate analysis;

[0020] 5) Incorporate the risk factor variables with P < 0.05 in the univariate analysis into a multivariate Cox regression model to construct a risk prediction model;

[0021] 6) Further, use the rms package of the R 3.6.1 software to establish a nomogram prediction model, calculate the risk score, and realize the visualization of the prediction model.

[0022] In the above construction method of the present invention, the sufficient number of case samples of acute leukemia patients undergoing hematopoietic stem cell transplantation refers to the quantity of sample data that meets the statistical requirements.

[0023] In the above construction method of the present invention, for the basic information, clinical symptoms, hematological indicators, and outcome indicators of patients meeting the inclusion criteria, the specific content of these terms is as follows:

[0024] Clinical data:

[0025] Demographic characteristics: gender, age.

[0026] Disease information: smoking history, drinking history, classification of hematopoietic stem cell source.

[0027] Hematological indicators:

[0028] (1) Blood routine items: white blood cells, red blood cells, red blood cell distribution width, hematocrit, platelets, hemoglobin, mean corpuscular volume.

[0029] (2) Liver function items: albumin, cholinesterase, total protein, alkaline phosphatase, alanine aminotransferase, direct bilirubin, γ-glutamyl transferase, total bilirubin, aspartate aminotransferase, lactate dehydrogenase.

[0030] (3) Renal function items: creatinine, uric acid, urea.

[0031] (4) Electrolyte items: potassium, calcium, chloride, sodium.

[0032] Patient outcome indicators:

[0033] For patients meeting the inclusion criteria, observe whether they develop oral mucositis, and divide the patients into the oral mucositis group and the non-oral mucositis group according to the outcome. The judgment criteria are as follows:

[0034] Oral mucositis group: symptoms such as erythema, ulcers (punctate, patchy, large-area rupture), erosion, oral pain, skin redness, gingival redness and pain, pseudomembrane or leukoplakia formation.

[0035] Non-oral mucositis group: without the above symptoms.

[0036] Furthermore, the data were sorted and statistically analyzed using SAS 9.4 described in step 4), including: categorical data and data were expressed as the number of cases and rates, and chi-square test or Fisher's exact test was used for between-group comparison; measurement data with normal distribution were described by mean plus or minus standard deviation, and t-test was used for between-group comparison; measurement data with skewed distribution were described by median and interquartile range, and Mann-Whitney U test was used for between-group comparison; linear mixed-effect model was used to calculate the fixed effects and random effects of the baseline levels and the slopes of changes over time of each repeated measurement index of the patients at four time points (before infusion, at the time of infusion, 1 - 3 days after infusion, and 4 - 7 days after infusion); the fixed effects of the model and the random effects of individual patients were used to estimate the baseline level of a certain index of each patient before infusion and the slope of change per unit time; Kaplan-Meier method was used to calculate the cumulative survival rate and draw the survival curve; ROC curve was used to find the optimal cut-off value of the index and then group, and log-rank test was used to compare the survival curves between groups.

[0037] Furthermore, for better fitting of the linear mixed effects model, logarithmic transformation is performed on non-normal data. If any two or three of the four time points of the patient are missing, the patient will not be included in the model.

[0038] Preferably, for the risk factors described in step 4), among them, age, baseline aspartate aminotransferase, alanine aminotransferase slope, baseline cholinesterase, lactate dehydrogenase slope, urea slope are positively correlated with the occurrence of oral mucositis, while baseline albumin, albumin slope, calcium slope are negatively correlated with the occurrence of oral mucositis.

[0039] Preferably, in step 5), for the variables or risk factors with P < 0.05 in the univariate analysis, among these factors, albumin slope ≤ -2.81, logarithmic slope of alanine aminotransferase > 0.29, baseline cholinesterase > 5900, urea slope > 0.48, calcium slope ≤ -0.242 are the risk factors for the occurrence of oral mucositis.

[0040] Optionally, in step 4), for the prediction model, the Bootstrap method is further used to resample repeatedly for internal validation of the prediction model to judge the stability of the model.

[0041] Optionally, in step 5), for the visualization of the prediction model, the Bootstrap method is further used to resample repeatedly to draw a calibration curve for internal validation of the nomogram model.

[0042] Preferably, in step 6), for the nomogram model, add each index to obtain the total score, and the probability of oral mucositis occurring at 1, 4, 7, 14, 21, and 30 days after the infusion of acute leukemia patients can be obtained corresponding to the nomogram. The results show that the prediction of oral mucositis occurring at 4 days, 7 days, 14 days, 21 days, and 30 days has a certain predictive value, and the model is relatively stable.

[0043] Preferably, the model constructed by the method for constructing the prediction model of the present invention as described above is used for predicting the risk of concurrent oral mucositis during hematopoietic stem cell transplantation in acute leukemia patients.

[0044] The construction method of the oral mucositis risk prediction model of the present invention involves analyzing risk factors by incorporating the basic information and hematological indicators of patients meeting the research design, screening prediction factors through the stepwise method and incorporating them into the variables of the prediction model, constructing the prediction model using multivariate Cox regression, and validating the model using the Bootstrap method. The research shows that this prediction model has good prediction efficacy, certain prediction value, and the model is relatively stable. The data source of this model has the characteristics of being easy to obtain, having objective indicators, and having less impact on patients, and is more suitable for front-line clinical medical staff to monitor oral mucositis, providing a basis for early detection, early intervention, and early treatment. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is a schematic diagram of the technical route of the research of the present invention;

[0046] Figure 2 It is a time - number distribution diagram of the occurrence of oral mucositis after reinfusion;

[0047] Figure 3 It is a time curve diagram of the cumulative incidence rate of oral mucositis after reinfusion;

[0048] Figure 4 It is a random effect fitting curve diagram of the linear mixed effect model of urea at each time point for acute leukemia patients;

[0049] Figure 5 It is a time curve diagram of the cumulative incidence rate of oral mucositis in acute leukemia patients undergoing hematopoietic stem cell transplantation calculated by the Kaplan - Meier method;

[0050] Figure 6 It is a time curve diagram of the cumulative incidence rate of oral mucositis in acute leukemia patients undergoing hematopoietic stem cell transplantation by log - rank test;

[0051] Figure 7 It is an ROC curve diagram of the occurrence of oral mucositis at 1, 4, 7, 14, 21, and 30 days after reinfusion for acute leukemia patients;

[0052] Figure 8 It is a nomogram of the risk prediction model for the occurrence of observation outcomes in acute leukemia patients;

[0053] Figure 9 It is a calibration diagram of the prediction model for the occurrence of oral mucositis 1 day after reinfusion in acute leukemia patients;

[0054] Figure 10 It is a calibration diagram of the prediction model for the occurrence of oral mucositis 4 days after reinfusion in acute leukemia patients;

[0055] Figure 11 It is a calibration diagram of the prediction model for the occurrence of oral mucositis 7 days after reinfusion in acute leukemia patients;

[0056] Figure 12 Calibration diagram of the prediction model for oral mucositis occurring 14 days after autologous hematopoietic stem cell transplantation in patients with acute leukemia;

[0057] Figure 13 Calibration diagram of the prediction model for oral mucositis occurring 21 days after autologous hematopoietic stem cell transplantation in patients with acute leukemia;

[0058] Figure 14 Calibration diagram of the prediction model for oral mucositis occurring 30 days after autologous hematopoietic stem cell transplantation in patients with acute leukemia. Detailed implementation manners

[0059] The following examples provide a more detailed description of the present invention. However, the following examples are provided only to assist in further understanding of the present invention and are not intended to limit the present invention. Those skilled in the art should understand that equivalent substitutions or corresponding improvements made to the content of the present invention still fall within the protection scope of the present invention.

[0060] Example 1 Construction of the oral mucositis risk prediction model

[0061] The technical route diagram of the construction method of the oral mucositis risk prediction model of the present invention is shown in Figure 1 .

[0062] 1. Research objects

[0063] This study was a retrospective cohort study. The research objects were all acute leukemia patients who underwent hematopoietic stem cell transplantation in the stem cell transplantation center of a tertiary hospital in Chongqing from January 1, 2013 to December 31, 2023. This study was approved by the Ethics Committee of the First Affiliated Hospital of Chongqing Medical University (Ethical review batch number: K2023-326). Since this study was a retrospective study, the informed consent of patients was waived after review by the ethics committee.

[0064] Inclusion criteria: 1. Acute leukemia patients who underwent hematopoietic stem cell transplantation; 2. Patients older than 18 years old; 3. Acute leukemia patients without oral infections, ulcers and other oral diseases before transplantation; 4. Patients who developed oral mucositis after autologous hematopoietic stem cell transplantation.

[0065] Exclusion criteria: 1. Patients diagnosed with non-acute leukemia; 2. Patients younger than 18 years old; 3. Patients with oral infections, ulcers and other oral mucositis symptoms before transplantation; 4. Patients who developed oral mucositis during the pretreatment period.

[0066] 2. Sample size calculation

[0067] This study was a model construction, and the empirical method for calculating the sample size of the model development cohort, that is, the principle of Events Per Variable (EPV), was used to estimate the sample size. The EPV principle refers to the ratio of the available sample size corresponding to each event (occurring outcome) in a clinical prediction model to the number of predictive variables. The empirical method ensures that at least 10 events occur for each predictive parameter considered for inclusion in the final predictive model equation. The sample size = number of variables × 10 / (1 - incidence rate). In this study, 10 variables were expected to be included in the final model. According to the literature review results, the incidence rate of oral mucositis during hematopoietic stem cell transplantation was expected to be between 50% and 80%. The calculated sample size was 200 - 500 cases. A total of 248 patients were included in this study, meeting the research requirements.

[0068] 3. Research Methods

[0069] Data Collection

[0070] This study adopted the method of retrospective data collection, continuously and dynamically collecting the basic information, clinical symptoms, hematological indicators, and outcome indicators of 248 acute leukemia patients during hematopoietic stem cell transplantation (from the day of pretreatment to the 30th day after hematopoietic stem cell reinfusion) from the electronic medical record system of a tertiary hospital's stem cell transplantation center in Chongqing.

[0071] Clinical Data

[0072] Demographic characteristics: gender, age.

[0073] Disease information: smoking history, drinking history, classification of hematopoietic stem cell source.

[0074] Hematological Indicators

[0075] (1) Blood routine items: white blood cells, red blood cells, red blood cell distribution width, hematocrit, platelets, hemoglobin, mean corpuscular volume.

[0076] (2) Liver function items: albumin, cholinesterase, total protein, alkaline phosphatase, alanine aminotransferase, direct bilirubin, γ-glutamyl transferase, total bilirubin, aspartate aminotransferase, lactate dehydrogenase.

[0077] (3) Renal function items: creatinine, uric acid, urea.

[0078] (4) Electrolyte items: potassium, calcium, chloride, sodium.

[0079] Patient Outcome Indicators

[0080] The outcome indicator was to observe whether oral mucositis occurred in patients meeting the inclusion criteria, and the patients were divided into the oral mucositis group and the non-oral mucositis group according to the outcome. The judgment criteria were as follows:

[0081] Oral mucositis occurred: symptoms such as erythema, ulcers (punctate, patchy, large - area ulceration), erosion, oral pain, skin redness, gingival redness and pain, pseudomembrane formation or leukoplakia formation, etc.

[0082] Oral mucositis did not occur: without the above - mentioned symptoms.

[0083] Statistical methods

[0084] SAS 9.4 was used for data sorting and statistical analysis. Categorical data were expressed as the number of patients and incidence rate, and the chi - square test or Fisher's exact test was used for between - group comparison. Measurement data with normal distribution were described by mean ± standard deviation, and the t - test was used for between - group comparison. Measurement data with skewed distribution were described by median and interquartile range, and the Mann - Whitney U test was used for between - group comparison. The linear mixed - effect model was used to calculate the fixed effects and random effects of the baseline levels and the slopes of the changes over time of each repeated - measurement index of patients at four time points (before infusion, at the time of infusion, 1 - 3 days after infusion, and 4 - 7 days after infusion). Then, the fixed effects of the model and the random effects of individual patients were used to estimate the baseline level of a certain index of each patient before infusion and the slope of the change per unit time. To better fit the linear mixed - effect model, logarithmic transformation was performed on non - normal data. If any two or three of the four time points of a patient were missing, the patient was not included in the model. The Kaplan - Meier method was used to calculate the cumulative survival rate and draw the survival curve. The ROC curve was used to find the optimal cut - off value of the index for grouping, and the log - rank test was used for comparing the survival curves between groups. Variables with P < 0.05 in the univariate analysis were included in the multivariate Cox regression model, and the step - by - step method was used for variable screening. The inclusion criterion was P < 0.05, and the exclusion criterion was P > 0.05. The area under the ROC curve was used to evaluate the prediction effect of the model. The Bootstrap method was used to resample 2000 times for internal validation of the prediction model to judge the stability of the model. The rms package of the R 3.6.1 software was used to establish a nomogram prediction model, calculate the risk score, and visualize the prediction model. The Bootstrap method was used to resample 2000 times to draw the calibration curve for internal validation of the nomogram model. When P < 0.05, the difference was considered to be statistically significant.

[0085] 4. Research results

[0086] 4.1 Overall situation of oral mucositis complicated during hematopoietic stem cell transplantation in patients with acute leukemia

[0087] A total of 248 patients with acute leukemia were included in this study. Among them, 145 developed oral mucositis after the reinfusion. The median follow-up time was 11.5 days. The main occurrence period was 5 - 10 days. The cumulative incidence after 15 days was 55.66%, and it increased slowly thereafter. The results are shown in Figure 2 、 Figure 3 and Table 1.

[0088] Table 1. Distribution table of the number of patients with oral mucositis after hematopoietic stem cell reinfusion in patients with acute leukemia

[0089]

[0090]

[0091] 4.2 Relationship between the risk of concurrent oral mucositis and clinical characteristics during hematopoietic stem cell transplantation in patients with acute leukemia

[0092] There was a statistically significant difference in the outcome index (whether oral mucositis occurred) among patients of different ages (P < 0.05). There were no significant differences in gender, smoking history, drinking history, disease type, and source of hematopoietic stem cells between the two groups (P > 0.05, see Table 2).

[0093] Table 2. Relationship between the risk of concurrent oral mucositis and basic data during hematopoietic stem cell transplantation in patients with acute leukemia

[0094]

[0095]

[0096] 4.3 Relationship between blood routine, liver function, kidney function, and electrolytes before hematopoietic stem cell reinfusion and the risk of oral mucositis in patients with acute leukemia

[0097] The cholinesterase and lactate dehydrogenase levels before hematopoietic stem cell reinfusion in patients with acute leukemia who developed oral mucositis were higher than those in patients who did not develop oral mucositis, and the difference was statistically significant (P < 0.05, see Table 3).

[0098] Table 3. Relationship between blood routine, liver function, kidney function, and electrolytes before hematopoietic stem cell reinfusion and the risk of oral mucositis in patients with acute leukemia

[0099]

[0100]

[0101] Note: / Fisher's exact test has no statistic.

[0102] 4.4 Relationship between the risk of oral mucositis and blood routine, liver function, kidney function, and electrolytes before and after hematopoietic stem cell reinfusion in patients with acute leukemia

[0103] The fixed effects and random effects of the baseline levels and the slopes of changes over time were analyzed using a linear mixed-effects model at the pre-reinfusion, during reinfusion, 1 - 3 days after reinfusion, and 4 - 7 days after reinfusion in patients with acute leukemia. Except for the lack of statistical significance in the changes of aspartate aminotransferase and lactate dehydrogenase at each time point, the other indicators changed over time (P < 0.05). Among them, white blood cell count, red blood cell count, red blood cell distribution width, hematocrit, platelet count, hemoglobin, mean corpuscular volume, albumin, cholinesterase, total protein, creatinine, uric acid, calcium, chloride, and sodium showed a downward trend over time, while alkaline phosphatase, alanine aminotransferase, direct bilirubin, γ-glutamyl transferase, total bilirubin, urea, and potassium showed an upward trend over time (see Table 4).

[0104] Taking urea as an example, the average baseline of urea in patients with acute leukemia before reinfusion was 4.090 mmol / L, and the average slope was 0.396 mmol / L, indicating that the slope of urea increased by an average of 0.396 mmol / L per unit change in unit time. The P values of the random effects of the baseline and the slope were <0.001 and 0.002 respectively, indicating that there were significant statistical differences in the baseline levels and slopes of urea among different patients. For example: Patient 1 did not develop oral mucositis, with a urea baseline level of 3.91 and a slope of 0.24. Patient 1 had an average increase in urea of 0.24 mmol / L per unit change in unit time, while Patient 2 developed oral mucositis, with a urea baseline level of 5.55 and a slope of 1.04. Patient 2 had an average increase in urea of 1.04 mmol / L per unit change in unit time (see Figure 4 ) Figure 4 It is the fitted curve of the random effect of the linear mixed-effects model of urea at each time point for patients with acute leukemia, where T1: before reinfusion, T2: during reinfusion, T3: 1 - 3 days after reinfusion, T4: 4 - 7 days after reinfusion.

[0105] Table 4. Fixed effects and random effects of various blood indicators before and after hematopoietic stem cell reinfusion in patients with acute leukemia

[0106]

[0107]

[0108] Note: a satisfies the normal distribution, and the data is logarithmically transformed.

[0109] 4.5 Relationship between the risk of oral mucositis and the baseline and change slopes of each indicator during hematopoietic stem cell transplantation in patients with acute leukemia

[0110] Before and after the reinfusion of hematopoietic stem cells in patients with acute leukemia, the baselines of aspartate aminotransferase, the slopes of alanine aminotransferase, the baselines of cholinesterase, the slopes of lactate dehydrogenase, and the slopes of urea were positively correlated with the occurrence of oral mucositis (P<0.05). While the baselines of albumin, the slopes of albumin, and the slopes of calcium were negatively correlated with the occurrence of oral mucositis (P<0.05, see Table 5).

[0111] Table 5. Relationship between the occurrence of oral mucositis during hematopoietic stem cell transplantation in patients with acute leukemia and the baselines and change slopes of various indicators

[0112]

[0113]

[0114] Note: a satisfies the normal distribution, and the data is logarithmically transformed.

[0115] 4.6 Survival curve of oral mucositis during hematopoietic stem cell transplantation in patients with acute leukemia

[0116] According to the optimal cut-off values of the baselines of aspartate aminotransferase, the baselines of albumin, the slopes of albumin, the slopes of alanine aminotransferase, the baselines of cholinesterase, the slopes of lactate dehydrogenase, the slopes of urea, and the slopes of calcium obtained from the ROC curve, they were divided into high and low groups. The cumulative survival rate was calculated using the Kaplan-Meier method, and the survival curves between groups were compared using the log-rank test (see Table 6 Figure 5 and Figure 6 ).

[0117] The cumulative incidence of oral mucositis was higher in the group with the logarithm of the baseline of aspartate aminotransferase > 3.25, the baseline of albumin ≤ 53, the slope of albumin ≤ -2.81, the logarithm of the slope of alanine aminotransferase > 0.29, the baseline of cholinesterase > 59, the logarithm of the slope of lactate dehydrogenase > -0.042, the slope of urea > 0.48, and the slope of calcium ≤ -2.42 than in its control group (P<0.05).

[0118] Table 6. Relationship between the risk of oral mucositis during hematopoietic stem cell transplantation in patients with acute leukemia and the baselines and change slopes of various indicators

[0119]

[0120]

[0121] 4.7 Influencing factors of oral mucositis during hematopoietic stem cell transplantation in patients with acute leukemia

[0122] Variables with P < 0.05 in the univariate analysis were included in the multivariate Cox regression model (variable assignments are shown in Table 7), and predictive factors were screened by the stepwise method: the inclusion criterion was P < 0.05, and the exclusion criterion was P ≥ 0.05. The results showed that albumin slope ≤ -2.81, alanine aminotransferase logarithmic slope > 0.29, cholinesterase baseline > 5900, urea slope > 0.48, and calcium slope ≤ -0.242 were risk factors for oral mucositis in patients with acute leukemia (see Table 8 for details).

[0123] Table 7. Variable Assignments

[0124]

[0125] Table 8. Results of the Cox Regression Model for Oral Mucositis during Hematopoietic Stem Cell Transplantation in Patients with Acute Leukemia

[0126]

[0127]

[0128] 4.8 Construction and Validation of a Risk Prediction Model for Oral Mucositis Complicated during Hematopoietic Stem Cell Transplantation in Patients with Acute Leukemia

[0129] According to the results of the multivariate Cox regression model analysis, a risk prediction model was constructed. The ROC curves for the occurrence of oral mucositis on the 1st, 4th, 7th, 14th, 21st, and 30th days after reinfusion are shown (see Figure 7 ). The area under the ROC curve of the risk prediction model for each time point was approximately 0.7. The Bootstrap method was used to resample 2000 times for internal validation of the prediction model. The results showed (see Table 9 for details) that the prediction of oral mucositis on the 4th, 7th, 14th, 21st, and 30th days had certain predictive value, and the model was relatively stable (the markers were all on the diagonal line).

[0130] Table 9. Evaluation Results of the Prediction Effect of the Risk Prediction Model

[0131]

[0132] Note: Calculated by the Bootstrap method, resampled 2000 times.

[0133] 4.9 Visualization of the Risk Prediction Model for Oral Mucositis Complicated during Hematopoietic Stem Cell Transplantation in Patients with Acute Leukemia

[0134] According to the factors in the multivariate Cox regression model, a nomogram model was constructed to visualize the prediction model. By adding each index, the total score was obtained, and the probability of oral mucositis occurring on the 1st, 4th, 7th, 14th, 21st, and 30th days after reinfusion of patients with acute leukemia could be obtained corresponding to the nomogram (see Tables 7, 11, andFigure 8 )。

[0135] After 2000 times of Bootstrap resampling verification, the calibration curve of this nomogram model is shown in ( Figures 9 - 14 ), and there is a good consistency between the probability of oral mucositis predicted by the risk prediction model and the actual probability of oral mucositis (see the red line on the diagonal).

[0136] Table 10. Score assignment of prediction factors in the nomogram model

[0137] Variable Score Albumin slope >-2.81 0 ≤-2.80 74 Alanine aminotransferase slope ≤0.29 0 >0.29 69 Cholinesterase baseline ≤5900 0 >5900 100 Urea slope slope ≤0.48 0 >0.48 60 Calcium slope >-0.242 0 ≤-0.242 61

[0138] Table 11. Total score of the nomogram model and the probability of oral mucositis in patients with acute leukemia after hematopoietic stem cell transplantation

[0139]

[0140] 5. Analysis of related risk factors

[0141] This study was a retrospective cohort study. A total of 248 patients with acute leukemia who underwent hematopoietic stem cell transplantation were included, and their basic information, clinical symptoms, hematological indexes and outcome indexes were retrospectively analyzed. After statistical analysis, this study found the risk factors of oral mucositis in patients with acute leukemia during hematopoietic stem cell transplantation. After Cox regression analysis, risk factors were included and a prediction model was constructed. This model has a certain predictive value, the model is relatively stable, and there is a good consistency between the probability of oral mucositis obtained and the actual probability of oral mucositis. The report is as follows:

[0142] 5.1 Correlation analysis between age and oral mucositis

[0143] We found that there was a statistical difference in age between the oral mucositis group and the non-oral mucositis group. With the increase of age in patients with hematopoietic stem cell transplantation, the risk of oral mucositis increased, and the incidence and severity of oral mucositis were higher in the young population. The metabolic level of young people is higher than that of middle-aged and elderly people, and the cell renewal cycle is fast. When cytotoxic drugs or antimetabolic drugs act on oral mucosal epithelial cells, if the induced apoptosis rate of cells is greater than the generation rate, oral mucositis will occur.

[0144] 5.2 Correlation analysis between changes in some liver function index levels and oral mucositis

[0145] 5.2.1 Correlation analysis between some liver metabolism enzymes and oral mucositis

[0146] This study shows that: (1) Before the hematopoietic stem cell reinfusion, the higher the baselines of cholinesterase and lactate dehydrogenase in patients with acute leukemia, the higher the risk of oral mucositis; (2) The greater the slopes of alanine aminotransferase and lactate dehydrogenase, the higher the degree of the risk of oral mucositis. When the logarithmic baseline of aspartate aminotransferase > 3.25, albumin baseline ≤ 53, albumin slope ≤ -2.81, logarithmic slope of alanine aminotransferase > 0.29, cholinesterase baseline > 59, and logarithmic value of lactate dehydrogenase > -0.042, the cumulative incidence of oral mucositis is significantly higher than that of its control group. Among them, albumin slope ≤ -2.81, logarithmic slope of alanine aminotransferase > 0.29, and cholinesterase baseline > 59 are risk factors for oral mucositis in patients with acute leukemia.

[0147] Aspartate aminotransferase, alanine aminotransferase, cholinesterase, lactate dehydrogenase, and albumin are of great significance in the monitoring of liver function. At different times, the liver function of patients with acute leukemia is affected by different factors. During the pretreatment period, it is mainly affected by the combined use of chemotherapeutic drugs; during the hematopoietic reconstruction period, antibiotic exposure can also lead to liver function impairment. The higher the cholinesterase and lactate dehydrogenase, the lower the liver function metabolism ability, the longer the damage time of chemotherapeutic drugs to the liver, the lower the drug clearance rate, and the increased incidence of oral mucositis. Therefore, front-line clinical medical staff should monitor the changes in liver function, and try to maintain the levels of some liver metabolism enzymes. When the values are lower than the risk values, comprehensive treatment should be carried out on patients as soon as possible to reduce the incidence of oral mucositis or relieve the symptoms of oral mucositis.

[0148] 5.2.2 Correlation analysis of albumin and oral mucositis

[0149] The lower the albumin baseline level and the smaller the albumin slope, the higher the risk of oral mucositis. Albumin is an important nutrient in the body, which reflects the nutritional status of the body to a certain extent. The albumin level is mainly affected by liver function, inflammatory response, nutritional status, endocrine factors, and genetic factors. In clinical practice, it is mainly used to monitor liver function, nutritional status, and evaluate the severity of diseases, etc. In cases of severe malnutrition, wasting diseases (such as malignant tumors, chronic wasting diseases, etc.) or intestinal malabsorption, etc., due to the liver toxicity of some chemotherapeutic drugs, the synthesis of albumin decreases, and the albumin level in the blood decreases. In patients with acute leukemia, factors such as long-term hypermetabolic state, reduced albumin intake due to drug factors, and impaired liver function lead to a low albumin level, which aggravates the occurrence of oral mucositis to a certain extent. Therefore, early prophylactic application of parenteral nutrition preparations to timely supplement energy and protein is of great significance for improving the nutritional status of patients with acute leukemia.

[0150] 5.2.3 Correlation analysis of urea and oral mucositis

[0151] This study shows that when the urea slope > 0.48, the cumulative incidence of oral mucositis increases, and the urea slope is positively correlated with the occurrence of oral mucositis. Urea is the main end product of human protein metabolism, mainly affected by protein intake, kidney function, and other factors (such as fever). In clinical work, it is mainly used for additional testing of kidney function, evaluation, or assisting in the diagnosis of diseases and detecting the therapeutic effect. The increase in urea is mainly affected by diet, weakened kidney function, and other factors. The higher the degree of urea increase before and after reinfusion, the higher the risk of the outcome. Most chemotherapy drugs (such as methotrexate, doxorubicin, cyclophosphamide, etc.) are mainly excreted from the body through the kidneys. In the case of renal insufficiency, the excretion of these chemotherapy drugs will be delayed, which may lead to higher drug concentrations in the blood and prolong the time of their damage to the body. Methotrexate belongs to antimetabolites, which inhibit the proliferation and proliferation of oral mucosal epithelial cells and directly induce oral mucositis, and is a risk factor for oral mucositis.

[0152] 5.2.4 Correlation analysis between calcium ion level changes and oral mucositis

[0153] This study proves that when the calcium slope ≤ -0.242, the cumulative incidence of oral mucositis increases, and the calcium slope is negatively correlated with the occurrence of oral mucositis. The faster the serum calcium ion decreases, the higher the probability of oral mucositis. Under normal circumstances, calcium is ingested orally, absorbed in the intestine, excreted and reabsorbed by the kidneys, and is affected by factors such as vitamin D, calcitonin, and diet. During the pretreatment stage, the patient's oral intake decreases, resulting in a reduction in calcium intake. After the pretreatment is completed, the patient's resistance decreases, and the risk of inflammatory infection doubles. After infection, the secretion of calcitonin increases, further reducing the loss of calcium ions in the body.

[0154] 6. Construction process and verification of the prediction model

[0155] Variables with P < 0.05 in the univariate analysis were included in the multivariate Cox regression model and assigned values respectively. Predictive factors were screened by the stepwise method, and the albumin slope, logarithmic slope of alanine aminotransferase, cholinesterase baseline, urea slope, and calcium slope were included in the prediction model for the occurrence of oral mucositis in patients with acute leukemia. According to the results of the multivariate Cox regression model analysis, a risk prediction model was constructed, and the ROC curves for the occurrence of oral mucositis at 1 day, 4 days, 7 days, 14 days, 21 days, and 30 days after reinfusion were calculated. After calculation, the area under the ROC curve of the risk prediction model at each time was about 0.7. After the model was constructed, internal verification of the prediction model was performed by repeated sampling 2000 times based on the Bootstrap method. The results showed that the prediction of the occurrence of oral mucositis at 4 days, 7 days, 14 days, 21 days, and 30 days had a certain predictive value, and the model was relatively stable.

[0156] The main achievements or technical effects of the present invention:

[0157] 1. A total of 248 patients with acute leukemia were included. Among them, 145 patients developed oral mucositis after hematopoietic stem cell infusion, and the overall incidence rate was 58.47%. The median follow-up time was 11.5 days. Oral mucositis mainly occurred in the period of 5 - 10 days. The cumulative incidence rate after 15 days was 55.66%, and then it increased slowly.

[0158] 2. There was a statistically significant difference in patient age between the non - oral mucositis group and the oral mucositis group (P < 0.05). There were no significant statistical differences in gender, smoking history, drinking history, disease type, and hematopoietic stem cell source between the two groups (P > 0.05).

[0159] 3. An albumin slope ≤ - 2.81, a logarithmic slope of alanine aminotransferase > 0.29, a cholinesterase baseline > 5900, a urea slope > 0.48, and a calcium slope ≤ - 0.242 were risk factors for oral mucositis in patients with acute leukemia. In addition, aspartate aminotransferase, alanine aminotransferase, cholinesterase, lactate dehydrogenase, and urea before and after hematopoietic stem cell infusion in patients with acute leukemia were positively correlated with the occurrence of oral mucositis (P < 0.05). While albumin and calcium ion levels were negatively correlated with the occurrence of oral mucositis (P < 0.05).

[0160] 4. The cumulative incidence rates of oral mucositis in the oral mucositis group with an aspartate aminotransferase logarithmic baseline > 3.25, an albumin baseline ≤ 53, an albumin slope ≤ - 2.81, a logarithmic slope of alanine aminotransferase > 0.29, a cholinesterase baseline > 59, a logarithmic lactate dehydrogenase > - 0.042, a urea slope > 0.48, and a calcium slope ≤ - 2.42 were all higher than those in their control groups.

[0161] 5. Based on aspartate aminotransferase, albumin, alanine aminotransferase, cholinesterase, lactate dehydrogenase, urea, and calcium, it has a certain predictive value for predicting the occurrence of oral mucositis in patients with acute leukemia at 4 days, 7 days, 14 days, 21 days, and 30 days during hematopoietic stem cell transplantation, and the model is relatively stable.

Claims

1. A method for constructing a risk prediction model for oral mucositis during hematopoietic stem cell transplantation in patients with acute leukemia, comprising the following steps: 1) Estimating sample size using the EPV principle of the model development cohort sample size calculation experience; 2) Collect enough samples of acute leukemia patients for hematopoietic stem cell transplantation and divide them into oral mucositis group and non-oral mucositis group; 3) Collect basic data, clinical symptoms, hematological indicators and outcome indicators of patients with acute leukemia undergoing hematopoietic stem cell transplantation; 4) SAS9.4 was used for data collation and statistical analysis, and risk factors were obtained through univariate analysis; 5) The risk factor variables with P < 0.05 in the univariate analysis were included in the multivariate Cox regression model to construct a risk prediction model; 6) Optionally, use the R3.6.1 software rms package to establish a nomogram prediction model, calculate the risk score, and visualize the prediction model.

2. The construction method as described in claim 1, wherein the SAS9.4 described in step 4) performs data collation and statistical analysis, including: categorical data and data are expressed as case numbers and rates, and inter-group comparisons are performed using chi-square test or Fisher's exact test; normally distributed quantitative data are described using mean plus or minus standard deviation, and inter-group comparisons are performed using t test; skewed distribution quantitative data are described using median and interquartile range, and inter-group comparisons are performed using Mann-Whitney U test; a linear mixed effects model is used to calculate the fixed effects and random effects of the baseline level and the slope of change over time of each repeated measurement index of the patient at four times (before re-infusion, during re-infusion, 1-3 days after re-infusion, and 4-7 days after re-infusion); the fixed effects of the model and the random effects of the individual patients are used to estimate the baseline level of a certain index of each patient before re-infusion and the slope of change over unit time; the Kaplan-Meier method is used to calculate the cumulative survival rate and draw the survival curve; the ROC curve is used to find the optimal cutoff value of the index and then group them, and the log-rank test is used to compare the survival curves between groups.

3. The construction method as described in claim 2 further comprises logarithmic transformation of non-normal data to better fit the linear mixed effects model, and if any two or three time points of the four time periods of the patient are missing, they are not included in the model.

4. The construction method according to claim 1, wherein the risk factor in step 4) is: Age, baseline aspartate aminotransferase, alanine aminotransferase slope, baseline cholinesterase, lactate dehydrogenase slope, and urea slope were positively correlated with the occurrence of oral mucositis, while baseline albumin, albumin slope, and calcium slope were negatively correlated with the occurrence of oral mucositis.

5. The construction method according to claim 1, in step 5), the variables or risk factors with P < 0.05 in the univariate analysis, among which albumin slope ≤ -2.81, alanine aminotransferase logarithmic slope > 0.29, cholinesterase baseline > 5900, urea slope > 0.48, and calcium slope ≤ -0.242 are risk factors for oral mucositis.

6. The construction method as claimed in claim 1, in step 4), further using the Bootstrap method to repeat sampling to perform internal verification of the prediction model to determine the stability of the model.

7. The construction method according to claim 1, in step 5), further using the Bootstrap method to repeat sampling to draw a calibration curve and perform internal verification on the nomogram model.

8. The construction method according to claim 1, in step 6), the nomogram prediction model adds each indicator to obtain a total score, and the corresponding nomogram can obtain the probability of oral mucositis in acute leukemia patients 1, 4, 7, 14, 21 and 30 days after re-infusion.

9. Use of the model constructed by the construction method according to claims 1 to 8 for predicting the risk of oral mucositis during hematopoietic stem cell transplantation in patients with acute leukemia.

Images