This study was approved by the Ethics Committee of West China Hospital, Sichuan University (approval no. 932 [2019]). The study objectives and methods were thoroughly explained to all participants, each of whom provided informed consent prior to participation. For minors (individuals aged <18 years), legal guardians provided consent on their behalf. In cases where participants lacked the capacity to consent due to illness, legal guardians signed the consent form.

The study was structured into three distinct components. Firstly, clinical data for patients were collected from the electronic medical record system, encompassing variables including age, gender, living environment, smoking and drinking habits, clinical symptoms, laboratory test results, sputum smear status, retreatment status, and multidrug-resistant tuberculosis (MDR-TB) status. Subsequently, a comparative analysis of clinical characteristics between patients with mild and severe TB was conducted. A molecular genetic susceptibility analysis for severe TB was then performed. The study comprised 558 patients with severe TB (case group) and 579 patients with mild TB (control group). DNA was extracted from peripheral blood samples obtained from all patients. Tag SNPs were selected based on genotype data from the Han Chinese population in Beijing, resulting in the selection of four specific ULK1 gene loci (rs9481, rs7138581, rs11616018, and rs1134574) for investigation in relation to TB severity. DNA extraction was conducted using the Tiangen Biotech DNA Extraction Kit (Beijing, China), and the concentration and purity of DNA in all samples were subsequently measured.

High-throughput SNP genotyping technology was employed to genotype the selected SNP sites. The allele frequency and genotype frequency distributions between mild and severe TB cases were compared to identify statistical differences. Four genetic models were analyzed: allelic, additive, dominant, and recessive. Finally, SNP loci identified as significantly associated with TB severity in the association analysis underwent clinical phenotype analysis, comparing differences in clinical symptoms, laboratory results, and other factors among different genotypes.

From January 2013 to December 2020, a total of 558 patients with severe TB and 579 patients with mild TB, aged 15–45 years and of Han ethnicity, were admitted to West China Hospital, Sichuan University. All included patients were unrelated individuals. The TB cases in this study were both clinically diagnosed and bacteriologically confirmed, adhering to relevant diagnostic criteria guidelines (14, 15). Patients with autoimmune deficiency syndrome, hepatitis B, diabetes, or history of organ transplantation were excluded.

Currently, universally accepted criteria for severe TB remain lacking. Based on a comprehensive literature review and guidelines from the World Health Organization (WHO), this study defines severe TB as meeting at least one of the following criteria (14–19):

Multi-system involvement: TB affecting two or more organ systems (e.g., pulmonary disease with concurrent meningeal or miliary dissemination).

Critical neurological involvement: Tuberculous meningitis, regardless of stage.

Disseminated disease: Hematogenous disseminated pulmonary TB.

Radiological severity: Lung cavities exceeding 4 cm in diameter, or bilateral pulmonary lesions involving >2/3 of the lung fields.

In the selection of tag SNPs, data from the Han Chinese population in Beijing, China were utilized. The selected tag SNPs adhered to a minimum allele frequency (MAF) of 5% and an r² value ranging from 0.8 to 1.0. Four specific tag SNP sites within the ULK1 gene, rs9481, rs7138581, rs11616018, and rs1134574, were included in this study. DNA extraction was performed using the Tiangen DNA Extraction Kit (Beijing, China) on 200 μL peripheral blood samples collected from all TB patients. All extracted DNA samples underwent assessment for concentration and purity, with samples considered qualified if their concentration exceeded 15 ng/μL and purity ranged from 1.8 to 2.0. Qualified samples were subsequently stored at −80°C. SNPscan™ multiplex SNP genotyping technology employed probes corresponding to distinct sequences at various SNP sites, thus enabling differentiation between alleles through detection of specific fluorescent signals.

Sample size was calculated using PASS 2023 software, with parameters set at α = 0.05 and statistical power (1-β) = 0.8. The statistical power for selected SNPs loci across various genetic models was subsequently evaluated using PASS 2023 software.

In comparing baseline data between mild and severe TB groups, as well as clinical phenotypic analyses of TB severity-associated loci, quantitative data are presented as mean ± standard deviation (SD), while qualitative data are expressed as proportions. The statistical significance of quantitative data was determined using t-tests or Mann–Whitney U tests, while chi-square tests were employed to compare proportions for qualitative data. Additionally, chi-square tests were used to ascertain statistically significant differences in allele and genotype frequency distributions between mild and severe TB cases. Four genetic models, allelic, additive, dominant, and recessive, were employed in the gene model analysis. Binary logistic regression was utilized to calculate associations between SNPs and TB severity, adjusting for confounding factors such as age and gender. To ensure the representativeness of the study sample, chi-square tests were conducted to confirm adherence of genotype frequencies to Hardy–Weinberg equilibrium (HWE). Finally, the online analysis tool SHEsis was employed to investigate linkage disequilibrium (LD) relationships among genetic markers and to assess haplotype composition (20).

In this study, 558 patients with severe TB and 579 patients with mild TB were observed. The patient screening process is shown in Figure 1. The mean age of patients in the severe TB group was 26.79 ± 8.17 years, slightly lower than 29.01 ± 8.20 years in the mild TB group. Smoking prevalence was notably higher in the severe TB group (24.6%) compared to the mild TB group (18.7%). Coughing and expectoration were the most common symptoms reported in both groups, affecting 47.5% of severe TB patients and 50.1% of mild TB patients. This was closely followed by fever, present in 39.1% of severe TB patients compared to 20.7% in the mild TB group. A higher proportion of patients in the severe TB group had a history of previous TB treatment (25.1%) compared to the mild TB group (14.5%). Additionally, positive sputum smear results were more prevalent in the severe TB group (29.6%) than in the mild TB group (19.0%). The prevalence of MDR-TB was slightly elevated in the severe TB group (3.4%) compared with the mild TB group (2.6%), as summarized in Table 1.

Four tag SNP loci within the ULK1 gene (rs9481, rs7138581, rs11616018, and rs1134574) were selected for analysis. The details of these SNPs are provided in Table 2. Logistic regression analysis indicated that the minor allele G of rs1134574 (A > G) significantly increased the risk of severe TB (OR = 23.034; 95% CI: 7.212–73.567; p < 0.0001). However, no significant correlation was found in the genotype frequency distributions (AA, AG, GG) or genetic model analyses (additive, dominant, and recessive models), with all p-values exceeding 0.05. After adjusting for age and gender as confounding factors in the logistic regression analysis, the minor allele G of rs1134574 remained significantly associated with increased severe TB risk (OR^a = 23.499; 95% CI: 7.339–75.249; p < 0.0001). Conversely, no significant associations were observed between the genotype frequency distributions of rs9481, rs7138581, or rs11616018 and severe TB in genetic model analyses, as presented in Table 3.

The HWE test was performed on both mild and severe TB groups, with p-values greater than 0.05 in all instances, indicating compliance with HWE and confirming the representativeness of the selected study population (Table 4).

LD analysis among the four selected tag SNPs in the ULK1 gene is depicted in Figure 2. Haplotype analysis was conducted using the SHEsis software platform, excluding haplotypes with frequencies below 0.03. Four major haplotypes (GCCG, GCTA, GGCA, and GGTA) were identified when examining associations between ULK1 SNPs and TB severity. Notably, the GCTA haplotype was significantly associated with a decreased risk of severe TB (OR^a = 0.746; 95% CI: 0.567–0.981; Pa = 0.036). Additional detailed results are presented in Table 5.

With a set significance level (α = 0.05) and statistical power (1 − β = 0.8), sample size calculations were performed using PASS 2023 software for odds ratios (ORs) of 1.5, 1.8, and 2.0, resulting in required case-group sample sizes of 484, 233, and 170, respectively (Figure 3). Additionally, we conducted sample power analyses for the four selected SNP loci across various genetic models using PASS 2023 software; results are presented in Table 6.

In the previous correlation analysis, a significant association was identified between the ULK1 gene SNP rs1134574 and TB severity. Specifically, the AA genotype was observed in 995 patients, the AG genotype in 136 patients, and the GG genotype in only 6 patients. Clinical phenotype analyses were performed for these three genotypes. A statistically significant difference was observed among genotypes regarding the proportion of patients experiencing tidal fever and night sweats (p = 0.045). However, no statistically significant differences were found among the genotypes for other clinical symptoms. Additionally, neutrophil percentages differed significantly across the AA, AG, and GG genotypes (p = 0.046; Table 7). No statistically significant differences were identified among genotypes in terms of sputum bacterial load, ICU admission rates, proportion requiring retreatment, length of hospital stay, rates of concurrent bacterial and fungal infections, or incidence of shock.