Tuberculosis (TB) is one of the leading causes of infectious disease related mortality worldwide. According to the “Global TB Report 2024”, the World Health Organization (WHO) estimated 10.8 million new TB cases and 1.25 million TB-related deaths in 2023 (World Health Organization, a). Notably, children and adolescents account for 12% of these new cases (World Health Organization, a). However, only 55% of these cases were reported to public health program (World Health Organization, a). The gap in case detection mainly stems from challenges in TB diagnosis. TB in children and adolescents often shows non-specific symptoms, such as fever, cough and weight loss (Inbaraj et al., 2025). Furthermore, the disease in this population is typically paucibacillary, and obtaining adequate sputum samples is difficult (Smith et al., 2023). Traditional diagnostic tools have poor performance in children, as less than 15% of children are sputum acid-fast bacilli smear positive. Mycobacterium tuberculosis (MTB) culture which is the “gold standard”, only yields 30%-40% positive results (Smith et al., 2023). In the new version of pediatric TB guidelines issued by WHO in 2022, it is recommended to use the Xpert MTB/RIF Ultra rapid molecular detection as the initial detection of pediatric TB (World Health Organization, b). However, the sensitivity of urine, feces and other non-sputum samples for the Xpert Ultra diagnosis of TB is still not high (Zar et al., 2019; Cox et al., 2022). Both tuberculin skin test (TST) and interferon gamma release assays (IGRAs) show reduced sensitivity in children with immature or compromised immune systems (Nolt and Starke, 2021). The lack of effective diagnostic tools leads to delayed treatment of TB. This delay of ATB diagnosis increases the risk of severe complications, such as miliary TB and tuberculous meningitis (Tedla et al., 2020). It also raises the chance of TB transmission among pediatric populations (Zhang et al., 2023). Globally, 96% of TB-related deaths in individuals younger than 15 years old occur in those who have not initiated treatment (World Health Organization, c; Bijker et al., 2024). Therefore, there is an urgent need for novel diagnostic tools.

In recent years, the neutrophil-to-lymphocyte ratio (NLR), a low-cost, easily calculable marker derived from routine complete blood counts (CBC), has emerged as a promising inflammatory biomarker for TB diagnosis. Multiple studies have validated its utility in distinguishing active TB (ATB) from latent TB infection (LTBI) or other infectious diseases (Abakay et al., 2014; Cursi et al., 2023). These findings based on adults have provided a foundation for applying of the NLR in TB infection. However, the application of NLR has several limitations. First, NLR lacks specificity, as it can be elevated in bacterial infections, malignancies and other conditions (Janka et al., 2023; Ueda et al., 2023). Second, NLR is susceptible to confounding factors like age, BMI and comorbidities (Tang et al., 2022). Third, the diagnostic cut-offs for NLR vary across different studies (Liu et al., 2022; Li et al., 2023). In addition, children and adolescents have immature immune systems and physiologically fluctuating NLR, so adult conclusions cannot be directly extrapolated. However, current pediatric studies on NLR and TB have present notable limitations. For instance, the study of Ma et al. included only 176 TB cases (Ma et al., 2024), and the study of Kissling et al. included just 12 TB cases (Sheng et al., 2025). Moreover, these studies did not adjust the confounding factors, such as age and immune status. The limitations of these results mean they are insufficient to meet clinical needs. Additionally, in clinical datasets, the raw NLR values often exhibit skewed distributions. This skewed distribution may reduce the stability of diagnostic cut-offs across different populations (Zahorec, 2021). Logarithmic transformation of NLR (logNLR) can enhance the consistency and reproducibility of biomarkers in inflammatory diseases (Tu et al., 2023; Man et al., 2025). However, the application of logNLR in the diagnosis of children and adolescents with ATB remains unexplored.

To address the unmet need for effective ATB diagnosis in children and adolescents, this study explored the diagnostic value of logNLR. The study enrolled 904 children and adolescents with ATB and 176 controls with non-ATB group infections (including bacterial pneumonia, non-tuberculous mycobacterial pulmonary disease, and parasitic infection). We collected demographic, clinical, and laboratory data, which included neutrophil and lymphocyte counts, and then applied receiver operating characteristic (ROC) curves, logistic regression analyses, restricted cubic spline (RCS) analyses, stratified analyses, and interaction analyses to assess the association between logNLR and ATB infection in this population. This study first sought to clarify the diagnostic performance of logNLR for children and adolescents with ATB. Moreover, we identified the potential non-linear relationship and critical thresholds between logNLR and ATB infection. Meanwhile, we further explored the stability of logNLR in different subgroups. This study intended to demonstrate logNLR as a simple, low-cost diagnostic biomarker for ATB in children and adolescents.

The interaction between innate and adaptive immune responses to MTB invasion is the potential mechanism that connects NLR to ATB infection (Mayer-Barber and Barber, 2015). Neutrophils serve as the first innate immune cells to infiltrate infection sites. When the host is challenged by external pathogens, neutrophils rapidly undergo chemotaxis, accumulate at the infected site, and contribute to the host’s defense against microbial invasion (Phillipson and Kubes, 2019). However, excessive neutrophils activation can induce tissue damage and trigger the release of pro-inflammatory cytokines (e.g., IL-1β, TNF-α), thereby disrupting immune homeostasis (Kolaczkowska and Kubes, 2013). Lymphocytes, particularly CD4+ and CD8+ T cells, are indispensable for adaptive immunity against MTB. During ATB infection, their counts often decline, which may favor MTB proliferation within the host (Wang et al., 2023). The NLR provides a more precise depiction of the dynamic balance between immune cells. The elevation of NLR indicates either increased neutrophil levels or reduced lymphocyte counts. Logarithmic transformation (logNLR) normalizes the skewed distribution of raw NLR values. As demonstrated by Zahorec in study on inflammatory biomarkers (Zahorec, 2021), this transformation enhances the reliability of NLR as a biomarker by minimizing variability associated with factors such as age and comorbidities.

Our study indicated that MLR (AUC = 0.6592) had slightly better diagnostic value than logNLR, while NMLR (AUC = 0.6260) and logNLR (AUC = 0.6214) demonstrated comparable diagnostic performance. In contrast, NAR (AUC = 0.5226) showed no effective discriminatory value. Notably, the ROC curves of MLR and logNLR exhibited highly similar trends, with close sensitivity and specificity performances. Although MLR showed slightly superior diagnostic efficacy, this advantage was minimal. Our study also confirmed that logNLR holds significant diagnostic value for ATB in children and adolescents even after adjusting for multiple confounding factors. The logistic regression results showed that in the fully adjusted model (Model 3), each 1-unit increase in logNLR was associated with a 38% higher risk of ATB. When logNLR was divided into two groups, the high logNLR group (Q2) still had a 49% higher risk of ATB than the low logNLR group (Q1). These findings align with prior pediatric studies. To discriminate between LTBI, ATB disease and other infectious diseases, Cursi et al. conducted a multicenter study involving 649 children, of whom 77 were diagnosed with ATB (Cursi et al., 2023). They study reported that NLR, with the cut-off of 1.5 and AUC of 0.72, had the sensitivity of 61% and specificity of 79%. Kissling et al. analyzed 389 children in Switzerland, 25 of whom had TB disease. They found that NLR was significantly elevated in children with TB disease compared to those with non-TB lower respiratory tract infections (nTB-LRTI). The cut-off value was 0.75, with an AUC of 0.82, sensitivity of 88%, and specificity of 71% (Kissling et al., 2023). Similarly, to compare severe TB (STB) and bacterial community-acquired pneumonia (CAP), Ma et al. conducted a study of 235 children, including 60 with STB, and found that NLR was an independent risk factor for STB. They also developed a predictive nomogram model with an AUC of 0.867, which demonstrated good discriminatory performance (Ma et al., 2024). However, unlike previous pediatric studies with small sample sizes, our study included 1,080 participants, including 904 ATB cases. The large sample size makes statistical results more scientific and reduces the bias caused by random variation in small samples. More importantly, our study accounted for key confounding variables (including age, gender, ethnicity, rural/urban, BCG vaccination status, TST results, exposure status and CD4+/CD8+ T cell counts) that had been overlooked in other pediatric investigations. For example, most newborns in China have been administered BCG vaccination, may influence immune markers including NLR (Guo et al., 2023). CD4+/CD8+ T cell counts reflect immune function, which is particularly important for children with immature immune systems (Sarfas et al., 2021). The inclusion of these covariates makes the association between logNLR and ATB more reliable.

Moreover, our study identified the non-linear relationship between logNLR and ATB infection risk and determined critical thresholds. The RCS analyses showed that the relationship between logNLR and ATB was non-linear (P for non-linearity=0.006), with a critical value of 0.9232. The breakpoint analyses further confirmed a breakpoint of 1.397. To the best of our knowledge, this non-linear pattern has not been reported in previous pediatric ATB studies and may be related to the physiological characteristics of pediatric immune systems. In children and adolescents with mild to moderate inflammation, elevated neutrophil counts or reduced lymphocyte counts drive a substantial rise in logNLR, exhibiting a strong linkage to the risk of ATB infection. The initial steep increase in risk up to logNLR=0.9232 may reflect the early activation of innate immunity and neutrophil recruitment in response to MTB infection (Phillipson and Kubes, 2019). However, when inflammation is severe (logNLR>1.397), the immune system may enter a state of “exhaustion” (Kolaczkowska and Kubes, 2013). The excessive neutrophil activation may cause tissue damage, and lymphocyte counts may be further suppressed, leading to irrelevance between logNLR and ATB infection. These thresholds (0.9232 and 1.397) can help clinicians stratify risk: children with logNLR<0.9232 may have a low ATB infection risk, while those with 0.9232≤logNLR ≤ 1.397 require priority evaluation for ATB.

Furthermore, stratified analyses revealed that logNLR as a biomarker for the diagnosis of ATB infection varies across subgroups. The results showed that logNLR was a significant positive predictor of ATB in children aged 0–7 and 15–17 years old, males, those with MTB exposure history, and those with CD4+ T cell counts>414 cells/μL or CD8+ T cell counts>238 cells/μL (all P<0.05). Among these, children aged 0–7 years old (OR: 1.91, 95% CI: 1.07-3.40) and adolescents aged 15–17 years old (OR: 3.09, 95% CI: 1.74-5.48) showed a stronger association, which may be attributed to immune responses. For young children, the number of CD4+ T cells in the adaptive immune system is low, and their overall immune function is immature. In young school-aged children, the numbers of CD4+ T cells and CD8+ T cells in the adaptive immune system are stable, and their immune function is relatively mature and stable. Adolescents are in the stage of immune system remodeling, where the activation state of CD4+ T cells is abnormal (e.g., increased proportion of HLA-DR+CD4+ T cells), and hormonal changes during puberty disrupt the immune balance (Donald et al., 2010; Basu Roy et al., 2019; Venturini et al., 2019). Notably, logNLR had no significant predictive effect in children with CD4+ T cell counts ≤ 414 cells/μL or CD8+ T cell counts ≤ 238 cells/μL (P>0.05). This suggests that in children and adolescents with impaired cellular immunity (e.g., due to malnutrition or underlying immune diseases), logNLR may not be a reliable biomarker—likely because severe immune suppression disrupts the balance between neutrophils and lymphocytes, weakening the correlation between logNLR and ATB infection. Clinicians should therefore combine other diagnostic tools (e.g., Xpert MTB/RIF) when evaluating ATB in immune compromised children. This study also explored the additive interaction between logNLR and other factors (CD4+ T cell counts, MTB exposure history and age). Children with both CD4+ T cell counts ≤ 414 cells/μL and MTB exposure had a much higher ATB infection risk (OR = 19.31) than those with either factor alone. The impaired cellular immunity (low CD4+ T cells) reduces host defense, and MTB exposure increases pathogen exposure. The combination of these two factors may enhance the susceptibility to ATB infection, thereby affecting the role of logNLR as a diagnostic marker for ATB. For those aged 0–14 years, immature immunity plus MTB exposure or low CD4+ T cells may further weaken anti-MTB capacity, amplifying ATB infection risk.

The data in this study were derived from a hospitalized population, and future research should expand to outpatient and primary-care settings to validate logNLR’s utility in these contexts, where early pediatric ATB screening is crucial. There is an urgent need, particularly in resource-limited areas, for a diagnostic tool that balances speed, accuracy, and operational feasibility to address the pediatric ATB diagnostic bottleneck (Basu and Chakraborty, 2025). Derived from routine complete blood count (CBC), logNLR is a biomarker that requires no specialized equipment or technical expertise, enabling rapid and low-cost preliminary assessment. This stands in contrast to advanced diagnostic tools, such as Xpert MTB/RIF Ultra or AI-based imaging, which remain difficult to implement in low-resource, high-burden settings due to their reliance on specialized infrastructure (Basu and Chakraborty, 2025). Thus, logNLR may serve as a simple, cost-effective triage tool. In primary-care settings, logNLR may serve as a triage tool to prioritize children with specific logNLR value ranges for further confirmatory testing, thereby reducing unnecessary referrals and accelerating treatment initiation. For outpatients with mild or nonspecific symptoms, logNLR may offer a low-cost screening option to rule out high ATB risk. However, this study has several limitations. First, it is a single-center retrospective study, which may introduce selection bias (e.g., all participants were hospitalized patients, excluding outpatients with mild ATB). Multicenter prospective studies are needed to validate the results in more diverse populations. Second, the non-ATB comparator group in this study did not include patients with viral respiratory infections, inflammatory non-infectious conditions, and healthy controls. Viral respiratory infections are highly prevalent among children and adolescents, and share similar clinical manifestations with ATB, rendering them a critical differential diagnosis in clinical practice. Inflammatory non-infectious conditions are characterized by systemic immune activation, which disrupts the balance between neutrophils and lymphocytes, and elevates NLR. This may result in false-positive outcomes when using logNLR for ATB diagnosis. Furthermore, healthy controls are essential for establishing a baseline logNLR range in the pediatric population, which enables the determination of whether the elevated logNLR is ATB-specific. Third, this study did not explore the dynamic changes in logNLR during anti-TB treatment. Therefore, it is unclear whether logNLR can be used to monitor treatment response. To address this limitation, prospective studies should track logNLR changes over treatment.

In conclusion, logNLR is a simple, low-cost, and effective biomarker for diagnosing ATB in children and adolescents, with its utility being particularly notable in males, 0–7 and 15–17 years old, those with MTB exposure history, and those with normal CD4+/CD8+ T cell counts. The critical thresholds (0.9232 and 1.397) provide a precise reference for clinical risk stratification. Although logNLR has limitations in immune compromised children, it still has important clinical value as an auxiliary diagnostic tool. Future research should focus on validating logNLR in external cohorts and multicenter populations, as well as exploring its combination with other biomarkers (e.g., NMLR, MLR) to further improve diagnostic accuracy. Additionally, it should investigate the utility of logNLR in monitoring treatment response.