Predictive value of bronchoalveolar lavage fluid interleukin-25, interleukin-33, and thymic stromal lymphopoietin levels for asthma development in children with recurrent wheezing Authors: Yu Zhao, Yaozheng Ling, Zhou Fu, Wenjing Zou Published: 2026 ABSTRACT AbstractObjectiveTo investigate the predictive value of interleukin-25 (IL-25), interleukin-33 (IL-33), and thymic stromal lymphopoietin (TSLP) levels in bronchoalveolar lavage fluid (BALF) for the development of asthma in children under 6 years of age with recurrent wheezing.MethodsA prospective cohort study was conducted from May 1 to September 30, 2024. Fifty children under 6 years of age with recurrent wheezing were enrolled as the wheezing group from the Department of Respiratory Medicine at Children’s Hospital of Chongqing Medical University, while 51 age-matched children with bronchial foreign bodies were recruited as controls during the same period. Levels of IL-25, IL-33, and TSLP in BALF were measured by enzyme-linked immunosorbent assay (ELISA). Children in the wheezing group were followed for one year and were subsequently categorized into an asthma group or a non-asthma group based on the follow-up outcomes. Integrating these clinical data with BALF cytokine levels, we constructed a predictive model for asthma development using receiver operating characteristic (ROC) curve analysis.ResultsUnivariate analysis revealed significant differences between the wheezing group and the control group in terms of clinical characteristics (the number of wheezing episodes, age, history of allergy, breastfeeding), peripheral blood eosinophil count, BALF parameters (nucleated cells, red blood cells, neutrophils, macrophages), and cytokine levels (IL-25, IL-33, TSLP) (P < 0.05). Of the 47 children in the wheezing group who completed the follow-up, 17 were diagnosed with asthma (asthma group) and 30 did not develop asthma (non-asthma group). Univariable analysis revealed significant differences between these two groups in terms of age, the number of wheezing episodes, levels of IL-25, IL-33, and TSLP in BALF (P < 0.05). Logistic regression analysis confirmed that these factors were significant predictors for the progression to asthma in children with recurrent wheezing (P < 0.05).ROC curve analysis demonstrated that the combination of the number of wheezing episodes with IL-25, IL-33, or TSLP levels yielded higher predictive values, with area under the curve (AUC) values of 0.90 (95% CI: 0.788-1.000), 0.946 (95% CI: 0.855-1.000), and 0.86 (95% CI: 0.698-1.000), respectively. These were superior to predictions based on wheezing episodes alone (AUC: 0.705) or individual cytokine levels (IL-25 AUC: 0.779; IL-33 AUC: 0.857; TSLP AUC: 0.841).ConclusionIn children under 6 years with recurrent wheezing, the levels of IL-25, IL-33, and TSLP in BALF are elevated and hold predictive value for the subsequent development of asthma, providing a reference for the early clinical identification of at-risk children.Keywords: Wheezing, Asthma, BALF, IL-25, IL-33, TSLP Introduction Recurrent wheezing is relatively common in preschool-aged children, and some patients may progress to persistent asthma. Early identification of children at high risk remains challenging because objective diagnostic tools-such as spirometry and bronchial provocation testing-are difficult to perform in children under six years of age. This highlights the urgent need for reliable predictive biomarkers. In recent years, the type 2 (T2) immune response has been confirmed as the core pathophysiological basis of asthma. A growing body of evidence indicates that airway epithelial cells serve not only as a physical barrier but also as pivotal sensors of environmental stimuli. Upon exposure to allergens (e.g., house dust mites, pollen), respiratory viruses, pollutants, or injury, the epithelium rapidly releases a set of epithelial-derived cytokines known as “alarmins”, primarily including IL-25, IL-33 and TSLP. These alarmins play a central role in the pathogenesis of asthma—particularly early-onset asthma—by acting as initial triggers and amplifiers of the inflammatory cascade [1]. Specifically, through distinct yet synergistic pathways, these three alarmins collectively establish the “epithelial-immune axis” that drives T2 inflammation in asthma [2]. TSLP primarily acts on dendritic cells, promoting the expression of co-stimulatory molecules and thereby driving native T-cell differentiation toward a T-helper 2 (Th2) cells phenotype, which helps establish allergic immune memory [3]. Concurrently, TSLP directly enhances the survival and function of group 2 innate lymphoid cells (ILC2s). IL-33 serves as the most potent activator of ILC2s, directly stimulating them to produce large quantities of IL-5 and IL-13 while also promoting the expansion of Th2 cells [4]. IL-25 also activates ILC2s, and upon binding to its receptor, can further amplify the response of ILC2s to IL-33, forming a positive feedback loop that perpetuates the inflammatory response [5]. Activated ILC2s and Th2 cells subsequently produce the effector cytokines IL-4, IL-5, and IL-13.Collectively, these downstream mediators contribute to airway eosinophilia, goblet cell hyperplasia and mucus hypersecretion, bronchial smooth muscle hyperresponsiveness, and the production of specific IgE, ultimately manifesting as the characteristic clinical and pathological features of asthma. Although numerous studies have documented elevated levels of IL-25, IL-33, and TSLP in BALF of asthma patients and have correlated their expression with disease severity in adults [6], their expression patterns and predictive value in young children with recurrent wheezing remain largely unknown. BALF analysis offers a direct means of assessing the local airway inflammatory milieu and is particularly suitable for young children undergoing bronchoscopy for clinical indications. To date, no study has systematically evaluated the predictive performance of BALF IL-25, IL-33, and TSLP for asthma development in children under six years of age with recurrent wheezing. Therefore, this prospective cohort study aimed to: (1) quantify the levels of IL-25, IL-33, and TSLP in BALF from children under six years with recurrent wheezing compared with bronchial foreign body controls; (2) determine whether these alarmins are independent predictors of subsequent asthma diagnosis during a one-year follow-up; and (3) evaluate whether combining alarmin levels with clinical characteristics (e.g., wheezing frequency) improves predictive performance. Our findings may provide a reference for early identification of at-risk children and inform targeted preventive strategies. Materials and methods Study design and participants This prospective cohort study was conducted from May 2024 to September 2024. Children under six years of age hospitalized for recurrent wheezing in the Department of Respiratory Medicine at Children’s Hospital of Chongqing Medical University were enrolled as the study population between May 1 and September 30, 2024. A control group was established using age-matched children with bronchial foreign bodies during the same period. All participants were screened according to predefined inclusion and exclusion criteria. The inclusion criteria of wheezing group: ①Acute-phase wheezing episodes at presentation, regardless of whether the trigger was infectious (e.g., viral respiratory infection) or non-infectious (e.g., allergen exposure, cold air, exercise);②Completion of pulmonary imaging and bronchoalveolar lavage(BAL) following admission;③Recurrent wheezing was defined as three or more wheezing episodes in the year prior to admission;④Children aged 0–6 years were included. Exclusion criteria comprised:①Premature birth, congenital airway malformations, pulmonary hypoplasia and other congenital disorders;②Structural or severe organic pulmonary diseases that may cause wheezing, such as bronchopulmonary dysplasia, interstitial lung disease, or cystic fibrosis;③Non-respiratory systemic diseases associated with wheezing (e.g., primary immunodeficiencies)n;④Severe pneumonia with substantial parenchymal destruction or chronic airway injury. The inclusion criteria of control group were as follows: ①Diagnosis of bronchial foreign body requiring bronchoalveolar lavage;②Children aged 0–6 years were included;③No evidence of pulmonary infection within the preceding 4 weeks. The study protocol was approved by the hospital’s Ethics Committee (Approval No.: 2022 − 163). Written informed consent was obtained from the guardians of all participants. A participant flow diagram is shown in Fig. 1. A participant flow diagram Data collection The following demographic and clinical data were systematically collected: sex, age, family history of atopy (asthma, rhinitis, food allergies, or atopic dermatitis in either parent), personal history of atopy, history of allergic rhinitis, breastfeeding history, and the number of wheezing episodes in the year prior to admission. Peripheral blood eosinophil percentage and BALF cellular profiles-including total cell count, nucleated cells, red blood cells, lymphocytes, neutrophils, eosinophils, macrophages, and columnar epithelial cells-were obtained from the Clinical Laboratory Department. These data were recorded for baseline comparison but were not included in the primary predictive analysis unless statistically significant in univariable screening. Bronchoalveolar lavage and cytokine measurement BAL was timely performed in all enrolled children following hospital admission. For children in the bronchial foreign bodies group, the initial BALF sample was collected from the contralateral (unaffected) lung during bronchoscopy. The lavage procedure followed a standardized protocol: each aliquot was instilled at 1 mL/kg, with a total volume not exceeding 20 mL per procedure, and a minimum single-aliquot recovery rate of > 40% was achieved. The recovered volume was recorded for each patient, and the median recovery rate was 52% (IQR: 44%–61%). Cytokine concentrations were not adjusted for recovery volume, as all analyses compared groups using the same procedure, and any dilutional effect would be non-differential. The collected BALF samples were immediately centrifuged at 112 ×g for 5 min at 4 °C. The supernatant was then separated and stored at − 80 °C until further analysis. For analysis, the supernatant was thawed at room temperature, and the concentrations of IL-25, IL-33, and TSLP in BALF were measured using ELISA. Each target molecule was measured in two independent plate runs on separate days. Within each plate run, all samples were tested in duplicate wells. The two duplicate readings were averaged to obtain a value for that run, and the two values from the independent runs were then averaged to determine the final concentration for each sample. If the coefficient of variation between the two runs exceeded 15%, the sample was re-assayed. The inter-assay coefficients of variation, calculated from quality control samples, were < 15% for IL-25, IL-33, and TSLP. All ELISA kits (48T/96T format; sensitivity: <0.1 pg/mL for IL-25 and IL-33, < 1.0 pg/mL for TSLP) were purchased from Quanzhou Ruixin Biological Technology Co., Ltd. (Quanzhou, China). According to the manufacturer’s specifications, the intra-assay coefficients of variation (CVs) were < 10% for all kits, and inter-assay CVs were < 15%,Spike-and-recovery rates ranged from 85% to 115%, confirming acceptable assay accuracy. All cytokine measurements were above the respective lower limits of detection for every sample; therefore, no imputation for values below the limit of detection was necessary. Follow-up All enrolled children underwent a one-year prospective follow-up to determine whether they ultimately developed asthma. The follow-up time point was 12 months after discharge, with follow-up methods including outpatient follow-up visits or telephone interviews. The follow-up items include: ① frequency of wheezing episodes in the child; ② whether asthma diagnosis-related tests (e.g., pulmonary function tests, bronchial dilation tests, and provocation tests) were performed; ③ whether asthma medications were administered. The diagnosis of asthma was made according to the Global Initiative for Asthma (GINA) 2024 criteria [7], which require both characteristic clinical symptoms (recurrent wheezing, shortness of breath, chest tightness, or cough, predominantly at night or upon waking) and objective evidence of variable airflow limitation (e.g., a positive bronchodilator test with an increase in FEV1 of ≥ 12% and ≥ 200 mL, or a positive bronchial provocation test with a decrease in FEV1 of ≥ 20%, or an average daily diurnal PEF variation > 10%). A detailed description of the GINA criteria is provided in the supplementary material. It should be noted that at the time of enrollment, none of the children in the wheezing group had been diagnosed with asthma, as they only presented with recurrent wheezing without meeting the diagnostic criteria for asthma at that time. Importantly, all cytokine levels and baseline clinical data were collected at the time of initial admission, before the follow-up outcomes were known, ensuring that baseline measurements were not biased by subsequent asthma diagnosis. During the one-year follow-up, 3 children in the wheezing group (n = 50) and 6 children in the control group (n = 51) were lost to follow-up, mainly due to failure to attend scheduled outpatient reexaminations or inability to contact by telephone despite multiple attempts. Among the 47 children who completed follow-up in the wheezing group, 17 were diagnosed with asthma (asthma group) and 30 did not develop asthma (non-asthma group). None of the children in the control group developed asthma; therefore, they were not included in the subsequent predictive analysis. Statistical analysis All statistical analyses were performed using R software (version 4.3.0). Continuous variables are presented as mean ± standard deviation (SD) for normally distributed data or median with interquartile range (Q1, Q3) for non-normally distributed data. Differences between two groups were assessed using the Mann-Whitney U test. Categorical data are expressed as numbers (n) and percentages (%), and compared using the chi-squared test or Fisher’s exact test, as appropriate. A two-sided P-value of less than 0.05 was considered statistically significant. Univariable analysis was first conducted for all potential factors. Variables that showed statistical significance (P < 0.05) were then included in a multivariable logistic regression model to identify independent predictors. The predictive performance of significant variables for asthma development was evaluated using ROC curves, and the AUC was calculated. Due to the limited number of asthma events (n = 17), no adjustment for multiple comparisons (e.g., Bonferroni correction) was applied, and internal validation (e.g., bootstrap optimism correction) was not performed; therefore, all findings should be interpreted as exploratory and hypothesis-generating. Results Demographic characteristics of the wheezing and control groups We recruited 50 children under six years with recurrent wheezing into the wheezing group and 51 age-matched children with bronchial foreign bodies into the control group. All enrolled children in this study resided in urban areas with no significant exposure to industrial pollution in their living environments. Therefore, environmental factors were not analyzed as statistical variables. Other demographic characteristics of the two groups are presented in Table 1. Compared with the control group, children in the recurrent wheezing group were younger (12.50 months vs. 20.00 months, P = 0.010), had a higher prevalence of family history of allergy (54% vs. 11.76%, P < 0.001), history of allergic rhinitis (54% vs. 11.76%, P < 0.001), and personal history of allergy (34% vs. 3.92%, P < 0.001), experienced more frequent wheezing episodes (3 times vs. 1 time, P < 0.001), and exhibited a higher percentage of peripheral blood eosinophils (0.02% vs. 0.01%, P = 0.042). However, no statistically significant differences were observed between the two groups in terms of BALF cell counts (total cell count, lymphocytes, eosinophils, and columnar epithelial cells) or sex distribution. Table 1Comparison of clinical parameters between the wheezing and control groupsVariablesTotal (n = 101)Wheezing Group (n = 50)Control Group (n = 51) P Sex, n(%)0.702 Male73 (72.28)37 (74.00)36 (70.59) Female28 (27.72)13 (26.00)15 (29.41)Age (months), M (Q₁, Q₃)18.00 (10.87, 28.00)12.50 (8.25, 30.75)20.00 (14.50, 25.00)0.010Family History of Atopy, n(%)< 0.001 No68 (67.33)23 (46.00)45 (88.24) Yes33 (32.67)27 (54.00)6 (11.76)History of Allergic Rhinitis, n(%)< 0.001 Absent68 (67.33)23 (46.00)45 (88.24) Present33 (32.67)27 (54.00)6 (11.76)Personal History of Atopy, n(%)< 0.001 Absent82 (81.19)33 (66.00)49 (96.08) Present19 (18.81)17 (34.00)2 (3.92)Breastfeeding, n(%)0.041 No29 (28.71)19 (38.00)10 (19.61) Yes72 (71.29)31 (62.00)41 (80.39)Number of Wheezing Episodes,M (Q₁, Q₃)3.00 (1.00, 3.00)3.00 (3.00, 4.00)1.00 (1.00, 1.00)< 0.001Peripheral Blood Eosinophils (%),M (Q₁, Q₃)0.02 (0.01, 0.03)0.02 (0.01, 0.04)0.01 (0.01, 0.03)0.042BALF Total Cell Count*106,M (Q₁, Q₃)1170.00(510.00, 2700.00)1275.00(625.50, 2615.00)1070.00(341.00, 2740.00)0.423BALF Nucleated Cells*106,M (Q₁, Q₃)800.00(325.00, 2130.00)1115.00(507.50, 2495.00)550.00(210.50, 1342.00)0.005BALF Red Blood Cells*106,M (Q₁, Q₃)100.00 (30.00, 565.00)45.00 (14.75, 195.00)280.00(69.00, 1059.00)< 0.001BALF Lymphocytes,M (Q₁, Q₃)18.00 (8.00, 65.70)38.55 (8.00, 105.41)15.00 (7.90, 37.00)0.061BALF Neutrophils,M (Q₁, Q₃)84.00 (40.00, 562.44)203.40(56.25,1392.70)65.00 (28.00, 173.00)0.008BALF Eosinophils,M (Q₁, Q₃)0.00 (0.00, 0.00)0.00 (0.00, 0.00)0.00 (0.00, 0.00)0.149BALF Macrophages,M (Q₁, Q₃)25.00 (9.00, 109.20)51.80 (14.00, 174.72)16.00 (8.00, 39.50)0.003BALFColumnar Epithelial Cells,M (Q₁, Q₃)16.00 (4.00, 78.00)19.90 (5.00, 85.50)15.00 (4.00, 64.26)0.731 Comparison of clinical parameters between the wheezing and control groups Number of Wheezing Episodes, M (Q₁, Q₃) BALF Total Cell Count*106, M (Q₁, Q₃) 1170.00 (510.00, 2700.00) 1275.00 (625.50, 2615.00) 1070.00 (341.00, 2740.00) BALF Nucleated Cells*106, M (Q₁, Q₃) 800.00 (325.00, 2130.00) 1115.00 (507.50, 2495.00) 550.00 (210.50, 1342.00) BALF Red Blood Cells*106, M (Q₁, Q₃) 280.00 (69.00, 1059.00) BALF Lymphocytes, M (Q₁, Q₃) BALF Neutrophils, M (Q₁, Q₃) BALF Eosinophils, M (Q₁, Q₃) BALF Macrophages, M (Q₁, Q₃) BALFColumnar Epithelial Cells, M (Q₁, Q₃) Comparison of cytokine levels between the wheezing group and the control group Cytokine level analysis revealed significantly elevated levels of all measured cytokines in the wheezing group compared to the control group. The median level of IL-25 was 23.81 (15.55, 30.96) pg/mL in the wheezing group versus 4.50 (2.13, 14.91) pg/mL in the control group (P < 0.001); the median level of IL-33 was 57.43 (32.73, 106.18) pg/mL versus 10.01 (6.47, 25.13) pg/mL in the control group (P < 0.001); and the median level of TSLP was 405.54 (381.06, 418.80) pg/mL versus 367.09 (352.60, 387.64) pg/mL in the control group (P < 0.001) (Fig. 2). BALF cytokine levels were compared between those in the wheezing group and the control group Univariable analysis of asthma development in children with wheezing A comparison of the asthma and non-asthma groups identified significant differences in several parameters, including age, number of wheezing episodes, and cytokine profiles. Children in the asthma group were significantly older (28.00; 13.00, 56.00months) than those in the non-asthma group (9.41; 8.03, 14.00months; P = 0.006). The distribution of wheezing episodes also differed significantly between the groups (P = 0.028). In the asthma group, the frequencies of wheezing episodes in the year prior to admission were: 3 episodes (35.29%), 4 episodes (35.29%), 5 episodes (17.65%), 6 episodes (5.88%), and 8 episodes (5.88%). In contrast, the non-asthma group distribution was: 3 episodes (76.67%), 4 episodes (13.33%), 5 episodes (3.33%), 6 episodes (3.33%), and 8 episodes (3.33%). Significant differences were also observed in BALF cytokine levels. The median IL-25 level was significantly higher in the asthma group (24.42) than in the non-asthma group (9.03; P = 0.011). Similarly, the asthma group showed elevated median levels of IL-33 (90.22 vs. 70.53; P = 0.002) and TSLP (413.04 vs. 276.73; P = 0.005) compared to the non-asthma group. Conversely, no statistically significant differences were found in other parameters, including BALF parameters (total cell count, nucleated cells, red blood cells, lymphocytes, neutrophils, macrophages, columnar epithelial cells, eosinophils), peripheral blood eosinophils, clinical history (breastfeeding, family history of atopy, allergic rhinitis, and personal atopy), or sex distribution (P > 0.05) (Table 2). Table 2Univariable analysis of asthma development in children with wheezingVariablesTotal (n = 47)Non-asthmatic group (n = 30)Asthma group (n = 17) P Sex, n(%)1.000 Male34 (72.34)22 (73.33)12 (70.59) Female13 (27.66)8 (26.67)5 (29.41)Age (months), M (Q₁, Q₃)13.00 (8.27, 30.50)9.41 (8.03, 14.00)28.00 (13.00, 56.00)0.006Family History of Atopy, n(%)0.560 No22 (46.81)15 (50.00)7 (41.18) Yes25 (53.19)15 (50.00)10 (58.82)History of Allergic Rhinitis, n(%)0.072 Absent22 (46.81)17 (56.67)5 (29.41) Present25 (53.19)13 (43.33)12 (70.59)Personal History of Atopy, n(%)0.242 Absent30 (63.83)21 (70.00)9 (52.94) Present17 (36.17)9 (30.00)8 (47.06)Breastfeeding, n(%)0.760 No18 (38.30)11 (36.67)7 (41.18) Yes29 (61.70)19 (63.33)10 (58.82)Number of Wheezing Episodes,M (Q₁, Q₃)0.028 329 (61.70)23 (76.67)6 (35.29) 410 (21.28)4 (13.33)6 (35.29) 54 (8.51)1 (3.33)3 (17.65) 62 (4.26)1 (3.33)1 (5.88) 82 (4.26)1 (3.33)1 (5.88)Peripheral Blood Eosinophils (%),M (Q₁, Q₃)0.02 (0.01, 0.05)0.02 (0.01, 0.05)0.02 (0.01, 0.03)0.535BALF Total Cell Count*106,M (Q₁, Q₃)1260.00(637.00, 2725.00)1150.00(536.00, 2052.50)1430.00(1010.00, 4250.00)0.245BALF Nucleated Cells*106,M (Q₁, Q₃)1090.00(480.00, 2665.00)1050.00(415.00, 2027.50)1400.00(730.00, 4100.00)0.268BALF Red Blood Cells*106,M (Q₁, Q₃)50.00 (20.00, 220.00)40.00 (10.75, 182.50)100.00 (30.00, 260.00)0.166BALF Lymphocytes,M (Q₁, Q₃)40.00 (8.00, 118.65)19.85 (8.25, 102.67)56.70 (8.00, 154.00)0.413BALF Neutrophils,M (Q₁, Q₃)194.40(48.50,1555.15)119.00(45.25,1154.10)262.80 (82.00, 3672.20)0.347BALF Macrophages,M (Q₁, Q₃)52.00 (15.00, 208.50)48.30 (14.00, 130.05)167.90 (25.00, 306.36)0.241BALF Columnar Epithelial Cells,M (Q₁, Q₃)22.00 (5.50, 86.60)32.50 (11.25, 86.90)13.00 (0.00, 60.00)0.164BALF Eosinophils,M (Q₁, Q₃)0.00 (0.00, 0.00)0.00 (0.00, 0.00)0.00 (0.00, 0.00)0.250IL-25, M (Q₁, Q₃)14.85 (7.96, 24.62)9.03 (5.27, 18.55)24.42 (14.07, 30.96)0.011IL-33, M (Q₁, Q₃)79.56 (60.63, 89.72)70.53 (34.16, 78.38)90.22 (83.36, 104.58)0.002TSLP, M (Q₁, Q₃)366.04(138.06, 405.54)276.73(116.97, 385.76)413.04 (357.37, 425.90)0.005 Univariable analysis of asthma development in children with wheezing Number of Wheezing Episodes, M (Q₁, Q₃) Peripheral Blood Eosinophils (%), M (Q₁, Q₃) BALF Total Cell Count*106, M (Q₁, Q₃) 1260.00 (637.00, 2725.00) 1150.00 (536.00, 2052.50) 1430.00 (1010.00, 4250.00) BALF Nucleated Cells*106, M (Q₁, Q₃) 1090.00 (480.00, 2665.00) 1050.00 (415.00, 2027.50) 1400.00 (730.00, 4100.00) BALF Red Blood Cells*106, M (Q₁, Q₃) BALF Lymphocytes, M (Q₁, Q₃) BALF Neutrophils, M (Q₁, Q₃) BALF Macrophages, M (Q₁, Q₃) BALF Columnar Epithelial Cells, M (Q₁, Q₃) BALF Eosinophils, M (Q₁, Q₃) 366.04 (138.06, 405.54) 276.73 (116.97, 385.76) Multivariable logistic regression analysis of asthma development in the wheezing cohort A multivariable logistic regression model was constructed to identify independent risk factors for asthma development. The dependent variable was the binary outcome of asthma development (yes/no). Independent variables included all factors that showed statistical significance in the Univariable analysis, which were appropriately coded prior to analysis. The regression analysis demonstrated that age, a history of more than three wheezing episodes, and elevated levels of IL-25, IL-33, and TSLP in BALF were significantly associated with asthma development (P < 0.05 for all), identifying them as independent risk factors for the progression from wheezing to asthma in this cohort (Table 3). Due to the limited number of asthma events (n = 17) relative to the number of covariates estimated (≥ 8), the events-per-variable ratio was approximately 2, which is below the recommended threshold of 10. Therefore, the following regression results should be interpreted as exploratory, and the coefficient estimates, particularly for categories with sparse data (e.g., 6 and 8 wheezing episodes), are unstable and require validation in larger cohorts. Table 3Multivariable logistic regression analysis of asthma development in the wheezing cohortVariableβ (Beta)SEZ scoreP valueOR95%CIAge (months)0.040.022.620.0091.051.01 ~ 1.08TSLP0.0101.960.0491.011.01 ~ 1.02IL-250.110.052.340.0191.111.02 ~ 1.22IL-330.060.032.290.0221.061.01 ~ 1.12Number of Wheezing Episodes 31.00 (Reference) 41.750.792.210.0275.751.22 ~ 27.14 52.441.241.970.04911.51.01 ~ 131.28 61.341.490.90.3663.830.21 ~ 70.64 81.341.490.90.3663.830.21 ~ 70.64 Multivariable logistic regression analysis of asthma development in the wheezing cohort Predictive value of wheezing frequency and BALF cytokine levels for asthma development in child ROC curves were plotted with the development of asthma in children with recurrent wheezing as the state variable (1 = asthma, 0 = non-asthma), and age, number of wheezing episodes, IL-25, IL-33, and TSLP levels as the test variables. The ROC analysis confirmed the significant predictive value of these factors. The optimal cut-off values determined by the ROC analysis were: age > 12.5 months, > 3 wheezing episodes, IL-25 > 21.348 pg/mL, IL-33 > 81.236 pg/mL, and TSLP > 408.269 pg/mL, identifying these thresholds as independent risk factors for asthma development in children with recurrent wheezing (Table 4). Furthermore, combining the number of wheezing episodes with IL-25, IL-33, or TSLP levels further enhanced the predictive performance, yielding AUCs (95% CI) of 0.9 (0.788–1.000), 0.879 (0.726–1.000), and 0.86 (0.698–1.000), respectively (Fig. 3). Table 4Predictive value of wheezing frequency and BALF cytokine levels for asthma development in childVariableAUC95%CISensitivitySpecificityOptimal Cut-off(original units / predicted probability)Youden’s IndexAge (months)0.7420.577–0.9080.670.8212.5000.49TSLP(pg/ml)0.8410.669–110.67408.2690.67IL-25(pg/ml)0.7790.607–0.9520.80.6221.3480.42IL-33(pg/ml)0.8570.710–1.0000.930.7981.2360.71Number of Wheezing Episodes0.7050.559–0.8500.770.653.50.41Number of Wheezing Episodes in combination with TSLP level0.860.698–1.0000.910.780.5770.69Number of Wheezing Episodes in combination with IL-25 level0.90.788–1.0000.760.920.670.69Number of Wheezing Episodes in combination withIL-33 level0.9460.855–1.0000.940.930.6260.87 Predictive value of wheezing frequency and BALF cytokine levels for asthma development in child Predictive performance of the ROC curve Of note, the reported AUCs were derived from the same sample used to build the predictive model. Due to the limited number of asthma events (n = 17), internal validation techniques such as bootstrap optimism correction or cross-validation could not be reliably performed. Consequently, these AUC estimates may be optimistic and should be interpreted as exploratory, warranting validation in independent external cohorts. For the combined models, the optimal cut-offs represent predicted probability thresholds derived from logistic regression models (range 0–1), rather than raw biomarker values. Specifically, predicted probabilities were obtained from logistic regression models that included both the number of wheezing episodes and the respective cytokine level as independent variables, and these probabilities were used as test variables in the ROC analysis. Discussion The key findings of this study are threefold. First, levels of IL-25, IL-33, and TSLP in BALF were significantly elevated in children under six years of age with recurrent wheezing compared with bronchial foreign body controls. Second, all three alarmins were independent risk factors for subsequent asthma development, with IL-33 showing the highest predictive value (AUC = 0.857). Third, combining the number of wheezing episodes with any of the three alarmins further improved predictive performance, achieving an AUC of up to 0.946. These findings suggest that BALF alarmins may serve as potential biomarkers for early identification of children at high risk of asthma. Clinical characteristics of children with recurrent wheezing The results of this study showed that, compared with the control group of children with bronchial foreign bodies, those in the recurrent wheezing group were younger, had higher proportions of family history of allergy, history of allergic rhinitis, and personal history of allergy, experienced more frequent wheezing episodes, and exhibited a significantly higher percentage of peripheral blood eosinophils. These clinical characteristics are largely consistent with the risk factors for wheezing reported in previous studies [8]. The core finding of this study is that the levels of IL-25, IL-33, and TSLP in BALF of children in the recurrent wheezing group were significantly higher than those in the control group with bronchial foreign bodies, suggesting a more active release of epithelial-derived alarmins and initiation of T2 inflammation locally in the airways of children with recurrent wheezing. In the control group of children with bronchial foreign bodies, although local airway irritation from the foreign body was present, there was no underlying chronic inflammation, and the BALF levels of alarmins were relatively low, further confirming the persistent activation of epithelial-derived airway inflammation in children with recurrent wheezing. For children under six years of age with recurrent wheezing, such elevated alarmin levels may reflect impaired airway epithelial barrier function and an underlying predisposition to T2 inflammation, thereby establishing an immunological foundation for the subsequent development of asthma. Risk of asthma development and clinical characteristics in children with recurrent wheezing In this study, the incidence of asthma among children with recurrent wheezing after one year of follow-up was 36.2%, which is generally consistent with previously reported rates in the literature. Previous studies have shown that approximately 30% to 40% of preschool children with recurrent wheezing ultimately develop asthma [9], indicating that not all children with wheezing will progress to persistent asthma and highlighting the urgent clinical need for effective risk stratification tools [10]. The present study found that children in the asthma group were significantly older than those in the non-asthma group (median age: 28.00 months vs. 9.41 months), suggesting that older children with recurrent wheezing may have a higher risk of developing asthma. This finding likely reflects the natural history of asthma onset—although wheezing may begin during infancy, a confirmed diagnosis of asthma often requires a period of observation and the accumulation of objective evidence. Furthermore, a significant difference was observed between the two groups in the distribution of wheezing episode frequency, with a markedly higher proportion of children experiencing > 3 wheezing episodes in the asthma group. Wheezing frequency is a key clinical indicator for assessing disease severity and risk of recurrence, and our results further support its clinical value in predicting the development of asthma. Notably, the absence of significant differences in peripheral blood eosinophils, family history of atopy, personal history of atopy, and allergic rhinitis between the asthma and non-asthma groups warrants comment. This finding may be attributable to the limited sample size (n = 17 events), which reduces statistical power to detect modest differences. Alternatively, in children under six years of age, these traditional atopic markers may be less mature or less predictive of long-term asthma progression compared with direct airway measures such as BALF alarmins, which reflect the local type 2 inflammatory milieu more directly. Predictive value of IL-25, IL-33, and TSLP for asthma onset The core finding of this study is that elevated levels of IL-25, IL-33, and TSLP in BALF are independent risk factors for the progression from recurrent wheezing to asthma in children. ROC curve analysis demonstrated that all three alarmins exhibited good predictive efficacy for asthma development, with IL-33 achieving an AUC of 0.857, which was higher than that of IL-25 (0.779) and TSLP (0.833). These findings suggest that among the three alarmins, IL-33 may have the optimal predictive value. As a member of the IL-1 family, IL-33 is the most potent activator of group ILC2s, directly stimulating them to produce large amounts of IL-5 and IL-13, thereby driving downstream type 2 inflammatory responses [11]. Previous animal studies have demonstrated that IL-33 is most highly expressed in the lungs during the neonatal period in mice, and that deletion of the IL-33/ST2 signaling pathway completely abrogates the development of asthma in neonatal mice, whereas it has only a limited effect in adult mice [12]. This age-dependent effect suggests that IL-33 may play a more critical role in the pathogenesis of childhood asthma than in adult asthma, which is highly consistent with our finding that IL-33 exhibited the optimal predictive value in children under six years of age with recurrent wheezing. Furthermore, as a nuclear factor, IL-33 is constitutively expressed in the nuclei of airway epithelial cells and is rapidly released upon epithelial cell injury or necrosis, a property that makes it a sensitive biomarker reflecting airway epithelial damage and inflammatory activation status [13]. TSLP also demonstrated good predictive efficacy for the development of asthma, with an AUC of 0.833. TSLP primarily acts on dendritic cells by promoting the expression of costimulatory molecules, thereby driving the differentiation of naïve T cells toward a Th2 phenotype and playing a key role in the establishment of allergic immunological memory [14]. Previous studies have indicated that TSLP and its receptor complex play important roles in the innate-adaptive immune response in childhood asthma, and animal experiments have confirmed that antagonizing TSLP significantly inhibits house dust mite-induced airway remodeling [15]. IL-25 exhibited relatively lower predictive efficacy (AUC = 0.779); however, as a member of the IL-17 family, it can amplify T2 inflammation by activating ILC2s and Th2 cells, and still holds important predictive value when acting in synergy with other factors. Clinical significance of joint prediction models The present study found that although individual indicators exhibited good predictive efficacy, the predictive performance was further enhanced when the frequency of wheezing episodes was combined with IL-25, IL-33, or TSLP, yielding AUC values of 0.900, 0.879, and 0.860, respectively. This finding aligns with the current trend in asthma biomarker research—namely, that a single biomarker often fails to fully capture disease heterogeneity, whereas an integrated strategy combining clinical characteristics with biomarkers offers significant advantages in identifying high-risk children [16]. In clinical practice, different disease entities or physiological conditions may present with similar wheezing symptoms, and relying solely on symptoms for prediction may suffer from insufficient specificity. Conversely, relying solely on a single biomarker may be influenced by various factors such as detection methods and inter-individual variability. The combination of both approaches allows for mutual complementation, thereby providing a more comprehensive and reliable assessment of the long-term risk of asthma development in affected children. In conclusion, as key initiators of type II inflammatory responses, IL-25, IL-33, and TSLP play essential roles in the progression from recurrent wheezing to asthma in preschool children. Measuring the levels of these cytokines in BALF, in combination with clinical characteristics such as the frequency of wheezing episodes, may help predict the risk of asthma development to a certain extent. However, to further enhance the objectivity and accuracy of clinical assessment, objective diagnostic tools should be integrated into routine clinical practice. For instance, fractional exhaled nitric oxide testing, spirometry, and impulse oscillometry can effectively aid in identifying the type of airway inflammation and assessing the degree of airflow limitation, thereby significantly improving the accuracy of disease classification and staging. Therefore, on basis of risk assessment using biomarkers and clinical features, the systematic incorporation of these objective diagnostic tools may facilitate earlier and more reliable identification of high-risk children, providing a more robust theoretical and practical foundation for implementing individualized intervention and management strategies. Advantage and disadvantage The innovations of our study are as follows. First, this study focused on a specific population of children under six years of age with recurrent wheezing, a group in whom the diagnosis of asthma is challenging due to a lack of objective predictive markers, thereby addressing an urgent clinical need. Second, we employed a prospective cohort design with a one-year follow-up, using a confirmed diagnosis of asthma as the outcome measure, which provides stronger evidence compared with cross-sectional studies. Third, we simultaneously measured three epithelial-derived alarmins (IL-25, IL-33, and TSLP) and systematically compared the predictive efficacy of individual indicators versus combined indicators, offering a comprehensive reference for clinical practice. Fourth, we integrated a clinical indicator (frequency of wheezing episodes) with biological indicators (cytokine levels) to establish a combined model with high predictive value, reflecting the principles of precision medicine. Despite the notable findings of this study, several limitations should be acknowledged. First, this was a single-center exploratory study with a small sample size and a limited number of children who developed asthma (n = 17). Given the exploratory nature and the low events-per-variable ratio (approximately 2) in the multivariable logistic regression model, the regression coefficients may be unstable and prone to overfitting. No adjustment for multiple comparisons (e.g., Bonferroni correction) was applied; therefore, reported P-values should be interpreted with caution. The reported AUCs were derived from the same dataset without internal validation (e.g., bootstrap optimism correction or cross-validation) due to the small event number, making the discriminative ability estimates likely optimistic. Thus, all findings should be considered hypothesis-generating rather than confirmatory, and external validation in larger independent cohorts is required.Second, the control group consisted of children with bronchial foreign bodies, which is a convenient comparator but has inherent limitations. The foreign body itself may induce local inflammation, and BALF was collected from the contralateral lung, which may not be entirely unaffected. Moreover, significant baseline differences in age, atopic history, and breastfeeding between the two groups limit direct comparability. Therefore, the control group comparison should be viewed as supporting biological plausibility rather than providing a definitive benchmark. Third, the one-year follow-up may not be sufficient to fully capture late-onset asthma cases. Fourth, the diagnosis of asthma in children under six years remains inherently challenging due to the lack of gold-standard objective tests. Although we strictly applied the GINA 2024 criteria, which require both characteristic clinical symptoms and objective evidence of variable airflow limitation, some degree of diagnostic misclassification cannot be completely excluded. Fifth, all children in the wheezing group underwent bronchoscopy as part of clinical evaluation, which may indicate more severe or refractory disease. The indication for bronchoscopy was not systematically recorded or adjusted for, and this may limit the generalizability of our findings to children with recurrent wheezing who do not require invasive investigation. Future studies in unselected populations are needed to confirm the external validity of our results. Based on the findings of this study, future research can be pursued in the following directions. First, multicenter prospective cohort studies should be conducted in collaboration with pediatric pulmonology departments from different regions and hospitals of various levels to expand the sample size, optimize and validate the current prediction model, and thereby improve its stability and generalizability. Second, long-term follow-up of the current study participants for an additional 3 to 5 years is warranted to document the occurrence of late-onset asthma cases, further evaluate the predictive value of cytokine levels for late-onset asthma, and supplement the long-term validity data of the prediction model. Third, the consistency of alarmin levels between non-invasive specimens—such as peripheral blood and exhaled breath condensate—and BALF should be explored, with the aim of developing more convenient detection methods. Acknowledgements Not applicable. Abbreviations Interleukin-33 Interleukin-25 Thymic stromal lymphopoietin Bronchoalveolar Lavage Fluid Bronchoalveolar Lavage Enzyme-linked Immunosorbent Assay Receiver Operating Characteristic Area Under the Curve World Health Organization Type 2 Innate Lymphoid Cells T-helper 2 Standard Deviation Authors’ contributions YZ and YZL were responsible for the conception and design of the manuscript, the statistical processing, analysis and interpretation of the results. YZ was responsible for conducting follow-up, observations of patients, carrying out the implementation and feasibility analysis of the study and the final writing of the manuscript. ZF and WJZ were responsible for the quality control and proofreading of the manuscript, as well as the overall supervision and management of the project. All authors read and approved the final manuscript. Funding The National Center for Clinical Medical Research in Children’s Health and Diseases Clinical Medical Research Youth Project (No. NCRCCHD-2021-YP-0X-11). Data availability The data sets analyzed during the current study are available from the corresponding author. Declarations Ethics approval and consent to participate The performed study’s protocol was following the principles of the declaration of Helsinki and approved by the Ethical Committee of Children’s Hospital of Chongqing Medical University (2022 − 163), and written informed consent was obtained from the guardians of all participants. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Zhou Fu, Email: fu_zhou79@126.com. Wenjing Zou, Email: 1016278469@qq.com. References Associated Data Data Availability Statement The data sets analyzed during the current study are available from the corresponding author.