Early life factors and genetic interactions in allergic rhinitis: a genome-wide study Authors: Shi Huang, Ziyan Zhang, Hongting Zhang, Ke Qiu, Tao Shu, Wei Xu, Juan Meng, Yu Zhao, Jianjun Ren Published: 2026 ABSTRACT AbstractBackgroundEarly life factors play an important role in the development and progression of allergic rhinitis (AR), but their relationship and underlying genetic mechanisms still remains unclear.MethodsUsing cross-sectional data from the UK Biobank (UKB) and West China Hospital (WCH), we assessed the associations between nine early life factors—birth weight, body size and height at age 10, multiple births, maternal smoking around birth, breastfeeding, long-term/recurrent antibiotic usage as a child or teenager (LTRAU), cesarean delivery, and maternal age—and the development of AR through multivariable logistic regression models. We also conducted a genome-wide gene-environment interaction study (GWGEIS) to explore the correlation between polygenic risk scores (PRS) derived from GWGEIS and AR risk.ResultsIn both UKB and WCH cohorts, LTRAU (UKB cohort: OR = 1.49, P < 0.001; WCH cohort: OR = 2.18, P < 0.001), and born by cesarean delivery (UKB cohort: OR = 1.26, P = 0.0014; WCH cohort: OR = 1.47, P = 0.008) were identified as independent risk factors for AR. Notably, the association between cesarean delivery and an increased risk of AR was confined to female offspring who developed AR before the age of 16. In contrast, the detrimental effects of LTRAU remained significant in offspring who developed AR after the age of 16. Additionally, GWGEIS identified an interaction between the ATE1 gene and LTRAU (rs4752620, P = 2.51E-8, β = -0.30). Furthermore, the PRS derived from GWGEIS results was significantly associated with the risk of AR (OR = 1.06, P = 0.038).ConclusionsThese findings enhance our understanding of how early life factors, particularly LTRAU, and genetic factors synergistically affect AR risk, offering valuable insights for prevention and treatment strategies.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12887-026-07052-6.Keywords: Allergic rhinitis, Asthma, Risk factors, Genome-wide by environment interaction studies Introduction Allergic rhinitis (AR) is a common IgE-mediated type I allergic disease that poses a significant challenge to global health [1]. The prevalence of AR is particularly high in the United States and Europe, affecting nearly 40% of the population in some regions [2]. In China, the prevalence of AR ranges from 9.8% to 23% [3]. The Developmental Origins of Health and Disease (DOHaD) theory suggested that early life events, such as prenatal and neonatal exposures, significantly affect the development of childhood diseases [4]. Previous studies have identified various early life exposure factors, including breastfeeding, having multiple siblings, advanced maternal age, born by cesarean delivery, long-term/recurrent antibiotic usage as a child or teenager (LTRAU), and maternal smoking during pregnancy, that might contribute to the risk of allergic diseases [5–8]. However, to date, most of the existing researches have focused primarily on asthma, with relatively few studies specifically addressing AR [9–11]. In addition, these investigations have predominantly concentrated on children and adolescents, leading to a gap in understanding the long-term effects of early life exposures on adult AR outcomes [12–15]. Genetic factors have also been implicated in the development of AR, with estimates of heritability ranging from 33 to 91% [16]. Despite the recent identification of numerous genes associated with AR and its complications, single nucleotide polymorphisms (SNPs) in these genes account for only 10% to 50% of the disease's heritability [17–19]. One potential explanation for this “missing heritability” is the role of gene-environment () interactions, as certain genetic variants may increase the risk of AR only in the presence of particular environmental exposures [20]. Despite the recognized importance of these interactions, causal research incorporating comprehensive genome-wide analyses in this area remains limited. This study aims to perform an observational investigation into the associations between nine early life factors and AR using data from the UK Biobank, with validation in the WCH cohort. Based on the results from these observational studies, we also conducted a genome-wide analysis to detect interactions between SNPs and LTRAU, as well as the association between polygenic risk scores (PRS) and the risk of AR. This analysis aims to assess the potential interactions among genetic factors, LTRAU, and the risk of developing AR, which may contribute to the formulation of early environmental intervention strategies to prevent AR. Methods Study design and participants The UK Biobank (UKB) cohort The UKB cohort (https://www.ukbiobank.ac.uk/) is a population-based cohort comprising approximately 500,000 participants, with their phenotypic and genotypic data being systematically documented. Participants included in this study were followed until June 22, 2022, with exclusions made for those missing key exposure variables. We examined nine early life exposure variables in this study, including body size at age of 10 (Field ID = 1687), height size at age of 10 (Field ID = 1697), multiple birth (Field ID = 1777), maternal smoking around birth (Field ID = 1787), breastfeeding status (Field ID = 1677), birth weight (Field ID = 20022), LTRAU (Field ID = 21067), born by caesarian section, and maternal age at delivery. Detailed assessments of these early life factors are presented in the Supplementary methods. AR was defined according to ICD-codes linked in hospital admission records: ICD-10 codes (J30.1, J30.2, J30.3, J30.4) and ICD-9 code (477), or self-reported doctor-diagnosed AR (Field ID = 22126). To avoid the influence of confounding factors and ensure the scientific rigor and accuracy of the analysis, patients with the following conditions closely related to AR were excluded from the main analyses: asthma, chronic sinusitis, nasal polyps, eczema, or dermatitis. (for detailed assessment information, see Supplementary methods). Similarly, in the subgroup analysis of patients with comorbid AR and asthma, we included only those with concurrent diagnoses of both conditions and excluded those with only a single diagnosis. The West China Hospital (WCH) cohort In a cross-sectional study based on the WCH database, we recruited 1043 parents of children aged between 8 months to 18 years old with AR, who visited the Allergic Reaction Center of West China Hospital between September and November 2023. Questionnaires were administered to assess their children’s early-life exposure factors (detailed information on the questionnaire is provided in Supplementary methods and Table. E1). Informed consents were obtained from all guardians of participants. Participants with asthma were excluded from this cohort. In the subgroup analyses, we included only those with both AR and asthma diagnoses. Covariates Since this study mainly focused on early life factors, we selected representative variables that do not change with age as covariates. In the UKB cohort, covariates included: birth year, gender, and ethnic background (White/Asian/others). In the WCH cohort, covariates included: demographic characteristics (age, gender), socioeconomic information (residential environment during childhood, household size, pet ownership during early life, maternal education and paternal education), and medical conditions (family allergic history). Statistical analyses Multivariable logistic regression analyses were performed to evaluate the association between early life factors and AR. In the UKB cohort, we used two models: Model 1: adjusted for birth year, sex, ethnic background; Model 2: included additional adjustments for the nine exposure variables, conducted simultaneously and mutually to minimize confounding. In the WCH cohort, due to the availability of extensive covariate data, a single model was used, adjusted for age, gender, family history of allergies, residential environment during childhood, household size, pets’ ownership during early life, parental education. This comprehensive adjustment ensures that confounding factors are minimized across both cohorts. To further assess the potential impact of multiple comparisons across the nine prespecified early-life exposure factors, we performed a sensitivity analysis using the Benjamini–Hochberg false discovery rate (FDR) procedure. Odds ratios (ORs) with 95% confidence intervals (CIs) and P values were calculated using multivariable logistic regression models. All statistical analyses were conducted using R3.6.3, and a significance level of P < 0.05 was considered statistically significant. Sensitivity analyses To obtain robust assessments of the associations between early life factors and AR, sensitivity analyses were conducted in the UKB cohort among participants with complete data on the seven variables with modest missing data (For more details, see Supplementary methods). Genotyping, imputation, and quality control Genotyping, quality control, and imputation were conducted on 488,377 UK Biobank participants with available genotype data. The specific steps for genotyping, imputation, and quality control are detailed in a previous study [21] and further elaborated in the Supplementary methods section. After quality control, a total of 9,014,180 variants were retained for the GWAS analysis, and 9,013,176 variants were retained for the GWGEIS analysis. Genome-wide gene-environment interaction analyses (GWGEIS) To explore the interaction between SNPs and LTRAU on AR risk, we performed a GWGEIS within the UK Biobank cohort. The genetic data passed quality control, and participants with LTRAU data were randomly divided into two subsets, AR-LTRAU1 and AR-LTRAU2. AR-LTRAU1 was utilized for genome-wide association analysis (GWAS) and GWGEIS using regression models incorporating SNP-environment interactions to enhance detection of genetic effects [22, 23]: where E was an environmental risk factor and G was a marker of interest, and ⊙ represented the product of them two. For both GWAS and GWGEIS, genome-wide significance was defined at p < 5 × 10⁻⁸, and suggestive associations were defined at p < 1 × 10⁻⁶, following recent GWAS conventions [24, 25]. More detailed information of GWGEIS is described in Supplementary methods. Function annotation and enrichment analyses The summary statistics of GWGEIS and GWAS were analyzed utilizing FUMA v1.3.5 software, including gene-based tests, mapping and functional annotations, gene prioritisation, and pathway enrichment (For more details, see Supplementary methods). Calculation of polygenic risk score In AR-LTRAU2, we employed PRScice to generate polygenic risk scores (PRSGxE) weighted according to GxE effects derived from AR-LTRAU1. Additionally, we constructed PRSAR in AR-LTRAU2 using SNP effects from the AR-GWAS of AR-LTRAU1, correcting for LTRAU. Subsequently, we used a model with AR as the outcome variable, incorporating PRSAR and the exposure factor LTRAU, to investigate the association between PRSGxE and AR. Detailed information is presented in the Supplementary Methods. Results Baseline characteristics A total of 379, 418 participants from the UKB were included in our analyses (Figure E1). The average age of the population was 56.5 years, with 53% identified as female, as shown in Table 1. Compared to those without incident AR, individuals with AR were more likely to be female, younger, and thinner or taller at age 10. Besides, individuals with AR were more likely to have been exposed to LTRAU, to have mothers aged 25–41 years, or to have been born between 1951 and 1971. Table 1Characteristics of participants with and without ARa in the UKB cohortbCovariateTotal (n = 379418)No (n = 328524)Yes (n = 50894)p-valueSexFemale202474 (53)174047 (53)28427 (56)< 0.001Male176944 (47)154477 (47)22467 (44)AgeMean (SD)56.5 (8.1)56.6 (8.1)55.3 (7.8)< 0.001Median (Min, Max)58 (39,73)58 (39,73)56 (40,70)Comparative body size at age 10average191945 (51)167572 (52)24373 (49)< 0.001Plumper58149 (16)50049 (16)8100 (16)Thinner121446 (33)103767 (32)17679 (35)Missing78787136742Comparative height size at age 10average203756 (55)178124 (55)25632 (51)< 0.001Shorter74592 (20)63840 (20)10752 (21)Taller93854 (25)80065 (25)13789 (27)Missing72166495721Part of a multiple birthNo364817 (98)315594 (98)49223 (98)< 0.001Yes8548 (2)7646 (2)902 (2)Missing60535284769Maternal smoking around birthNo232716 (71)200532 (71)32184 (72)< 0.001Yes94718 (29)82287 (29)12431 (28)Missing51984457056279Breastfed as a babyNo78840 (27)67601 (27)11239 (27)0.21Yes211087 (73)181380 (73)29707 (73)Missing89491795439948Birth weightNormal159535 (77)135254 (76)24281 (79)< 0.001Low20588 (10)17774 (10)2814 (9)High28039 (13)24217 (14)3822 (12)Missing17125615127919977Long-term/recurrent antibiotic usage as child or teenagerNo108776 (86)86981 (88)21795 (81)< 0.001Yes17050 (14)12048 (12)5002 (19)Missing25359222949524097Born by caesarian sectionNo127036 (97)99659 (97)27377 (97)< 0.001Yes3499 (3)2590 (3)909 (3)Missing24888322627522608Mother age on date of delivery ≤ 18Y3134 (2)2758 (2)376 (2)< 0.00119Y-24Y47544 (32)40566 (32)6978 (29)25Y-41Y97614 (65)81233 (65)16381 (68) ≥ 42Y1207 (1)981 (1)226 (1)Missing22991920298626933Ethnic backgroundWhite358280 (95)310120 (95)48,160 (95)< 0.001Asian8377 (2)7323 (2)1054 (2)Other11266 (3)9779 (3)1487 (3)Missing14951302193Birth Year1934–1950195708 (52)174392 (53)21316 (42)< 0.0011951–1960117417 (31)99206 (30)18211 (36)1961–197166293 (17)54926 (17)11367 (22)aUKB UK BiobankbAR allergic rhinitis The bold items in this table indicates that p-values that are less than 0.05 Characteristics of participants with and without ARa in the UKB cohortb aUKB UK Biobank bAR allergic rhinitis The bold items in this table indicates that p-values that are less than 0.05 In the WCH cohort, the average age of participants with AR was 9.6 years, with 62.1% being male. As shown in Table 2, compared to those without AR, participants with AR were younger and had a higher likelihood of exposure to LTRAU. They were also thinner at age 10, mostly born via caesarian section, had mothers aged ≥ 26 years, had a family history of allergies, lived in urban areas during childhood, and did not have pets (cats or dogs) during early life. Besides, their parental education levels were more likely to be between 12 and 15 years. Table 2Characteristics of participants with and without ARa in the WCH cohortbCovariateTotal (n = 944)No (n = 424)Yes (n = 520)p-valueComparative body size at age 10average274(62.6)150(68.8)124(56.4)0.0259Plumper44(10)19(8.7)25(11.4)Thinner120(27.4)49(22.5)71(32.3)Missing506206300Comparative height size at age 10average288(65.8)154(70.6)134(60.9)0.0948Shorter82(18.7)34(15.6)48(21.8)Taller68(15.5)30(13.8)38(17.3)Missing506206300Part of a multiple birthNo914(96.8)407(96)507(97.5)0.2591Yes30(3.2)17(4)13(2.5)Missing///Maternal smoking around birthNo939(99.5)420(99.1)519(99.8)0.18Yes5(0.5)4(0.9)1(0.2)Missing///Breastfed as a babyNo198(21)87(20.5)111(21.3)0.8179Yes746(79)337(79.5)409(78.7)Missing///Birth weightNormal863(93.2)382(92.5)481(93.8)0.1878Low29(3.1)11(2.7)18(3.5)High34(3.7)20(4.8)14(2.7)Missing18117Long-term/recurrent antibiotic usage as child or teenagerNo531(56.2)286(67.5)245(47.1)< 0.001Yes413(43.8)138(32.5)275(52.9)Missing///Born by caesarian sectionNo442(46.8)232(54.7)210(40.4)< 0.001Yes502(53.2)192(45.3)310(59.6)Missing///Mother age on date of delivery< = 25Y204(21.7)107(25.5)97(18.7)0.03626-35Y644(68.7)276(65.9)368(70.9)> 35Y90(9.6)36(8.6)54(10.4)Missing651AgeMean (SD)9.6(6.1)9.7(6.2)9.5(6)0.2831Median [Min, Max]8[0.8, 18]8[0.8, 17]8[1, 18]SexFemale364(38.6)180(42.5)184(35.4)0.0314Male580(61.4)244(57.5)336(64.6)Family allergic historyNo480(50.8)281(66.3)199(38.3)< 0.001Yes464(49.2)143(33.7)321(61.7)Residential environment during childhoodUrban area788(83.5)312(73.6)476(91.5)< 0.001Countryside156(16.5)112(26.4)44(8.5)Number in household during childhoodMean (SD)4.2(1.1)4.3(1.2)4.1(1.1)0.0909Median [Min, Max]4[1, 10]4[1, 8]4[2, 10]Cats or dogs during early lifeNo770(81.6)326(76.9)444(85.4)0.0011Yes174(18.4)98(23.1)76(14.6)Maternal education< 12Y147(15.6)99(23.3)48(9.2)< 0.00112-15Y632(66.9)241(56.8)391(75.2)≥ 16Y165(17.5)84(19.8)81(15.6)Paternal education< 12Y140(14.8)94(22.2)46(8.8)< 0.00112Y-15Y671(71.1)265(62.5)406(78.1)≥ 16Y133(14.1)65(15.3)68(13.1)aWCH West China HospitalbAR allergic rhinitis The bold items in this table indicates that p-values that are less than 0.05 Characteristics of participants with and without ARa in the WCH cohortb aWCH West China Hospital bAR allergic rhinitis The bold items in this table indicates that p-values that are less than 0.05 Multivariable logistic regression analyses investigated associations between early life factors and AR incidence in two independent cohorts To investigate the associations between nine early life factors and the incidence of AR, we conducted multivariable logistic regression analyses in both the UKB and WCH cohorts. To assess potential multicollinearity among the exposure variables, we conducted a Pearson correlation analysis. The results revealed no significant linear correlations among the exposure variables (Figure. E2.). Consequently, all nine exposures were included in Model 2 for further adjustment in the multivariable analyses. The results from the multivariable analyses adjusted by Model 2 showed that abnormal height (taller: 95%CI: 1.02–1.15, P = 0.0067, OR = 1.08; shorter: 95%CI: 1.07–1.22, P < 0.001, OR = 1.14) or body size (plumper: 95%CI: 1.03–1.19, P = 0.0046, OR = 1.11; thinner: 95%CI: 1.11–1.24, P < 0.001, OR = 1.17) at age 10, LTRAU (95%CI: 1.4–1.59, P < 0.001, OR = 1.49), born by caesarian delivery (95%CI: 1.09–1.46, P = 1.26, OR = 1.26) and maternal age at delivery > 18 years were significant risk factors for AR. In contrast, maternal smoking around birth (95%CI: 0.87–0.98, P = 0.012, OR = 0.93, Model 2) and being part of a multiple birth (95%CI: 0.68–0.99, P = 0.035, OR = 0.82, Model 2) showed potential protective effects against the incidence of AR (Fig. 1, Table E2,3.). The association between nine early-life exposure factors and allergic rhinitis in the UKB and WCH cohorts. The UKB Cohort, The UK Biobank Cohort; The WCH Cohort, The West China Hospital Cohort Similar results were observed in the WCH cohort, where thinner body size at age 10, born via caesarian section, and LTRAU were associated with a higher risk of developing AR (Fig. 1, Table E2,4.). Additionally, the FDR-corrected results were highly consistent with the primary analysis and did not materially alter the overall interpretation (Table E5.). Furthermore, sensitivity analysis was conducted in the UKB cohort by including participants with fewer missing values for all seven exposure factors. The results remained consistent with those generated from the main analyses (Table. E6.). Stratification analyses based on gender and age at diagnosis of AR to examine the associations in the UKB Cohort To further examine the mediating effects of gender and age at diagnosis on the observed associations, we conducted stratification analyses based on these two covariates (Table. E7-11.). Specifically, regarding age at diagnosis, AR patients were separated into early-onset (age at diagnosis ≤ 16 years) and late-onset (age at diagnosis > 16 years) subgroups with reference to the acknowledged definitional standard of asthma24. Among female participants, consistent results were observed. LTRAU was significant risk factor for both early and late-onset AR. Participants born by caesarian section were at a significantly higher risk of developing early-onset AR (Table. E11.). Among male participants, no significant association was observed between caesarian delivery and AR. Whereas LTRAU was significantly associated with both early- and late-onset AR (Table. E11.). Subgroup analyses in patients with comorbid AR and asthma investigated early life factor associations across two independent cohorts Since AR and asthma are both atopic diseases and often co-occur, we conducted subgroup analyses to explore the associations between early-life factors and the comorbidity of AR with asthma. Participants with a diagnosis of both AR and asthma were included, while those with only a single diagnosis were excluded. In the UKB cohort, information on asthma status was available for 343, 856 AR patients (Table. E12.). Compared to our main analyses (Fig. 1, Table E2,3.), LTRAU, and born by caesarian section were higher in AR patients with asthma (Fig. 2, Table. E13-16.). In contrast to the findings in the main analyses (Fig. 1, Table E2,3.), no significant association was observed between maternal age at delivery and the occurrence of comorbid AR and asthma (Fig. 2, Table. E13-16.). The association between nine early-life exposure factors and allergic rhinitis with comorbid asthma in the UKB and WCH cohorts. The UKB Cohort, The UK Biobank Cohort; The WCH Cohort, The West China Hospital Cohort In the WCH cohort, asthma status was available for 502 AR patients (Table. E17.). Compared to the UKB cohort results, in the WCH cohort, born by cesarean section did not have a statistically significant effect on comorbid AR and asthma (Fig. 2, Table E13,18.). GWAS implicated risk variants and related functional pathways in AR We then conducted a GWAS of AR, summarizing lead SNPs with P-values < 1 × 10−6 in Table E19. The Manhattan plots illustrating the results are shown in Fig. 3. Our GWAS identified nine significant loci. The lead SNPs for these loci included, for example, rs2212434 (P = 2.21E−10, β = 0.13, nearest gene: C11orf30), rs113412244 (P = 4.84E−8, β = 0.12, nearest gene: IL18R1), rs4766578 (P = 7.67E−9, β = −0.12), and rs9915135 (P = 2.47E−8, β = 0.13). Subsequent functional annotations mapped these SNPs to seven genes: IL1RL1, IL18R1, TLR1, RPS26, SMARCE1, LRRC32, and IL33. Besides, gene-based tests identified four genes with genome-wide significance for AR, corrected using the Bonferroni method (P = 2.63E−6): IL18R1 (P = 5.06E−7), SMARCE1 (P = 5.16E−7), KRT222 (P = 1.93E−7&3.77E−7), and TLR1 (P = 2.80E−7) (Table E20.). These variants were found to be involved in functional pathways, including the positive regulation of NAD+ nucleosidase activity and the response to bacterial lipoproteins (Table E21.). Manhattan plot for the genome-wide association study (GWAS) and genome-wide gene-environment interaction study (GWGEIS) of allergic rhinitis in individuals of European ancestry from the UK Biobank (UKB) cohort. GWAS,genome-wide association study;GWGEIS,genome-wide gene-environment interaction study GWGEIS revealed gene-environment interaction effects on the development of AR Among the numerous environmental factors associated with AR, we focused on LTRAU for further GWGEIS due to its significant association in both the UKB and WCH cohorts. The GWGEIS identified a significant candidate G × E interaction locus: an intronic SNP in the ATE1 gene on chromosome 10 (rs4752620, P = 2.51E-08, β = −0.30) (see Table E19.). The Manhattan plots displaying the interaction results are presented in Fig. 3. Subsequent functional annotation linked this SNP to the ATE1 gene. While gene-based tests(Table E20.)did not yield significant findings, gene set enrichment analysis provided weak evidence (P < 0.05) suggesting a potential involvement of artery development in the observed interaction (Table E21.). More details are provided in Table E19-21. To further examine the association between the polygenic risk score for PRSGxE, calculated based on the GWGEIS results, and AR, we conducted multivariable logistic regression analyses. Specifically, in age- and gender- adjusted analyses, when incorporating PRSAR (calculated based on the GWAS results of AR) and the exposure factor LTRAU into the model, PRSGxE was found to be significantly associated with AR (OR = 1.06, P = 0.038) (Table 3). This finding reinforces the gene-environment interaction effects on the development of AR. Table 3The associations between LTRAUa, PRSARb, PRSGxEc and ARd by multivariable logistic regressionExposureOutcome: ARCase (n = 5,356)Control (n = 68,862)OR (95% CI)p-valueLTRAU984/43729825/590371.3 (1.21, 1.4)< 0.001PRSAR2454/290234655/342071.2 (1.13, 1.27)< 0.001PRSGxE2605/275134504/343581.06 (1, 1.12)0.038Multivariable logistic regression analyses were adjusted for age and gender. Additionally, when assessing the associations between PRSGxE and AR, PRSAR and the exposure factor LTRAU were also included in the adjustment model. The bold items in this table indicates that p-values that are less than 0.05aLTRAU long-term/recurrent antibiotic usage as a child or teenagerbPRSAR Polygenic Risk Scores calculated based on the GWAS results of ARcPRSGxE Polygenic Risk Scores calculated based on the genome-wide gene-environment interaction study results of ARdAR allergic rhinitis The associations between LTRAUa, PRSARb, PRSGxEc and ARd by multivariable logistic regression Multivariable logistic regression analyses were adjusted for age and gender. Additionally, when assessing the associations between PRSGxE and AR, PRSAR and the exposure factor LTRAU were also included in the adjustment model. The bold items in this table indicates that p-values that are less than 0.05 aLTRAU long-term/recurrent antibiotic usage as a child or teenager bPRSAR Polygenic Risk Scores calculated based on the GWAS results of AR cPRSGxE Polygenic Risk Scores calculated based on the genome-wide gene-environment interaction study results of AR dAR allergic rhinitis Discussion Our study provides new insights into how early life factors and genetic interactions contribute to AR development. Notably, we identified LTRAU and cesarean delivery as common risk factors for AR across both European and Chinese populations, corroborating previous findings in European cohorts while extending these insights to Chinese populations. Additionally, our GWGEIS results highlight the role of the ATE1 gene in gene-environment interactions, a novel finding that has not been reported in previous AR-related studies. This study not only confirms the importance of early life environmental exposures but also underscores the need for further research into the genetic mechanisms underlying AR susceptibility. Our findings suggests that LTRAU is a common risk factor for AR across both European and Chinese populations, aligning with previous studies predominantly focused on European populations [26, 27]. This consistency underscores the potential for LTRAU's adverse effects to persist into adolescence and adulthood. Given the detrimental consequences of antibiotic misuse, it is imperative for healthcare professionals and national healthcare planners to devise strategic approaches for the effective management of antibiotics. In routine medical practice, it is also crucial to maintain awareness of the principle of prudent antibiotic use. Similarly, cesarean delivery emerged as a shared risk factor for AR in both populations [28, 29], likely due to the lack of exposure to maternal vaginal microbiota during birth, which may impede immune system maturation and disrupt the balance of Th1 and Th2 lymphocytes critical for immune regulation [30]. Our stratification analysis further revealed that cesarean delivery significantly increases the risk of AR in female offspring under the age of 16, suggesting that early-life microbiota exposure may be particularly impactful during this developmental window [31]. China is among the countries with the highest cesarean section rates globally, with a prevalence of 34.9% [32].In light of the emerging association between cesarean delivery and the subsequent development of AR in offspring, the decision to undergo this surgical procedure should be made with more caution, particularly when the infant is female. This study observed a slight inverse association between maternal smoking during the perinatal period and the risk of AR in offspring (OR = 0.93, 95% CI 0.87–0.98, P = 0.012). However, this unexpected direction of association should be interpreted with caution. Several factors may underlie this finding. First, the effect size was small, suggesting a marginal association that might not reflect a true biological effect. Second, the exposure definition in our study was broader, encompassing both late-pregnancy and early postnatal smoking, which could lead to exposure misclassification. Third, recall and survivor bias in the UK Biobank—where participants were middle-aged adults recalling exposures from decades earlier—might have attenuated positive associations or even reversed their direction. Fourth, unmeasured socioeconomic and environmental confounders may have influenced both maternal smoking behavior and AR risk, introducing residual confounding. Previous studies have reported either no significant association or even an inverse relationship between maternal smoking during the perinatal period and the risk of AR in offspring [33–35]. Therefore, the inverse association observed here is more likely attributable to methodological and confounding effects rather than a genuine protective effect. Given the modest effect size and inconsistency with existing evidence, this result should be interpreted cautiously, and further validation in prospective cohorts and multi-omics studies is warranted. In the present study, multiple births were identified as a potential protective factor against AR. The observed association may be interpreted in the context of early-life immune modulation and microbial exposure. Previous cohort data indicate that children from multiple births show lower incidence of hay fever and other allergic diseases—potentially reflecting the “sibling/multiple-birth effect,” in which increased early microbial and antigenic exposure enhances immune regulation and tolerance development [36, 37]. Additionally, larger in-utero or perinatal immune challenges inherent to multiple births may accelerate maturation of regulatory T-cell populations or shift the Th1/Th2 balance toward less allergy-prone phenotypes, although direct mechanistic studies in AR remain limited [38]. These hypotheses underline the need for future mechanistic investigations combining immune phenotyping and microbiome profiling in cohorts of multiple versus singleton births to clarify the biological basis of this protective association. In this study, we explored the genetic basis of AR through GWAS and identified eight genes significantly associated with the disease using functional genomics approaches. The combined effects of these genes may influence the onset and development of AR by modulating various aspects of the immune system, such as inflammatory responses, immune tolerance, and cytokine secretion. For instance, IL1RL1 and IL33 have been shown to play crucial roles in allergic diseases by activating Th2-type immune responses, while IL18R1 may enhance immune activation by promoting inflammatory responses [39]. Toll-like receptor 1 (TLR1), an important pattern recognition receptor, could alter the local immune environment by sensing pathogen invasion, thereby affecting the clinical manifestation of AR [40]. The role of LRRC32 (GARP) in regulatory T cells suggests that it may be critical in maintaining immune tolerance and suppressing excessive immune responses [41]. The eight genes identified in this study through GWAS have been previously confirmed to be closely associated with AR, further validating the robustness of our findings. The GWGEIS in our study identified a significant gene–environment interaction locus within the ATE1 gene, which encodes arginyltransferase 1, a key enzyme mediating N-terminal arginylation. ATE1 has been implicated in angiogenesis, tumor cell differentiation, neuroregulation, protein degradation, cellular stress responses, and inflammatory signaling pathways [42–44]. Arginylation mediated by ATE1 can modulate NF-κB, MAPK, and mitochondrial oxidative stress pathways, thereby influencing macrophage activation and tissue repair [45–47]. These pathways are also critically involved in chronic inflammation and epithelial barrier regulation in AR [48–50]. Collectively, these findings suggest that ATE1 may represent a novel susceptibility gene for AR with cross-population relevance. Future studies should validate these results through expression analyses in clinical samples and disease-relevant animal models to elucidate the functional role of ATE1 in AR pathogenesis. In our study, we also calculated PRS for AR (PRSAR) and GWGEIS (PRSGxE). Specifically, in age- and gender-adjusted analyses, we noticed that including PRSAR and LTRAU in the model demonstrated a significant association between PRSGxE and AR. This indicates that the identified gene-environment interaction variants play a crucial role in AR development and may represent some novel risk loci for AR beyond those detected by conventional GWAS, thereby addressing some aspects of missing heritability [20]. This study has several notable strengths. First, it represents one of the initial efforts to explore the influence of LTRAU and genetic factors on AR development at a genome-wide level. Second, the study population encompasses individuals of all ages, rather than being limited to children and adolescents. Third, by integrating data from both European and Chinese populations, the robustness of our findings is enhanced. However, there are limitations of our study. First, reliance on self-reported early-life exposure data may introduce recall bias. To mitigate this, we conducted sensitivity analyses in the UKB cohort and using a cross-sectional design in the WCH cohort for younger participants. Second, the absence of detailed information on important factors such as breastfeeding duration and maternal smoking history limits our ability to explore dose–response relationships. Third, the lack of genetic data from the WCH cohort restricts our ability to directly compare genetic risks between the European and Chinese populations. In conclusion, through multivariable logistic regression analyses of data from the UKB and the WCH cohorts, we identified several factors influencing AR risk. Notably, maternal smoking during the perinatal period and multiple births were found to be protective against AR. In contrast, LTRAU, cesarean delivery, thinner body size at age 10, and older maternal age at childbirth emerged as significant risk factors. Our GWAS identified nine candidate loci associated with AR, while our GWGEIS highlighted the interaction with the ATE1 gene, confirming a significant association between polygenic risk scores for PRSGxE and AR. These findings underscore the importance of integrating genetic and environmental factors in AR prevention and management strategies. Supplementary Information Supplementary Material 1. Supplementary Material 2. Supplementary Material 3. Supplementary Material 4. Acknowledgements Not applicable Abbreviations Allergic rhinitis UK Biobank West China Hospital Genome-wide gene-environment interaction study Polygenic risk scores Long-term/recurrent antibiotic usage as a child or teenager Single nucleotide polymorphisms Authors’ contributions MJ, RJJ, and ZY conceived and designed the study. HS, ZZY, and ZHT acquired, analyzed, and interpreted the data. HS and ZZY drafted the manuscript, while MJ, RJJ, XW, QK, and ZY critically revised it for important intellectual content. HS, ZZY, and ST performed the statistical analysis. All authors provided final approval of the manuscript for submission and publication. Funding This work was supported by the Chengdu Science and Technology Bureau (2022-YF05-01387-SN), the Science and Technology Department of Sichuan Province [K.Q., 2024NSFSC1510;J.J.R., 2024YFHZ0335], the Sichuan Cadre Health Care Committee (2022–113), and the National Clinical Research Center for Geriatrics, West China Hospital, Sichuan University (Z2023LC002). Data availability The UKB data are available in a public, open-access repository. This study utilized the UK Biobank resource under application numbers 69718 and 80787. The UK Biobank data can be accessed upon application to the UK Biobank (www.ukbiobank.ac.uk/). To protect participants' identities, the WCH cohort data will not be made publicly available. Interested parties can contact Prof. Ren (Jianjun.Ren@scu.edu.cn), who will provide the complete raw data. Declarations Ethics approval and consent to participate The UK Biobank study was approved by the National Health Service National Research Ethics Service (ref. 21/NW/0157, Application number: 80787). The study based on the WCH cohort was approved by the ethics committee of West China Hospital, Sichuan University [2023(N0.1488)].Written informed consent for participation was obtained from all participants prior to enrollment. For participants younger than 16 years of age, written informed consent for participation was obtained from their parents or legal guardians. All procedures involving human participants were performed in accordance with the ethical standards of the institutional and/or national research committee and with the Declaration of Helsinki. Consent for publication For the UK Biobank cohort, patient consent for publication was not required, as the data are publicly accessible. For the WCH cohort,written informed consent for publication of the study results was obtained from all participants or, for participants younger than 16 years of age, from their parents or legal guardians. 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. Shi Huang, Ziyan Zhang and Hongting Zhang contributed equally to this work. Juan Meng, Yu Zhao and Jianjun Ren contributed equally to this work. Contributor Information Juan Meng, Email: mjmelinda@163.com. Yu Zhao, Email: yuzhao@wchscu.edu.cn. Jianjun Ren, Email: Jianjun.Ren@scu.edu.cn. References Associated Data Supplementary Materials Supplementary Material 1. Supplementary Material 2. Supplementary Material 3. Supplementary Material 4. Data Availability Statement The UKB data are available in a public, open-access repository. This study utilized the UK Biobank resource under application numbers 69718 and 80787. The UK Biobank data can be accessed upon application to the UK Biobank (www.ukbiobank.ac.uk/). To protect participants' identities, the WCH cohort data will not be made publicly available. Interested parties can contact Prof. Ren (Jianjun.Ren@scu.edu.cn), who will provide the complete raw data.