Tuberculosis (TB) remains one of the deadliest infectious diseases worldwide, with an estimated 10 million new cases and 1.5 million deaths annually (World Health Organization, 2024). Despite significant progress in TB control, critical gaps remian in both diagnosis and treatment monitoring (World Health Organization, 2024). Early and accurate diagnosis, coupled with effective monitoring of treatment response, is essential to reducing TB transmission and improving patient outcomes (Teibo et al., 2024).

Current TB diagnostic methods each face significant challenges. Acid-fast bacilli (AFB) smear microscopy, though widely available, suffers from variable and often low sensitivity (Huang et al., 2022). This limitation is particularly pronounced in paucibacillary TB, including child TB (Shingadia and Novelli, 2003), extrapulmonary TB (Bajracharya et al., 2022), or HIV coinfected TB (Bajracharya et al., 2022). Mycobacterial Culture is the gold standard for TB diagnosis, mainly including solid and liquid Culture. Mycobacteria Growth Indicator Tube (MGIT), a key liquid culture technique. Mycobacterial Culture requires specialized biosafety facilities and has a prolonged culture time (6–8 weeks), limiting its utility in routine clinical practice (Huang et al., 2022). Although GeneXpert MTB/RIF has revolutionized TB diagnostics with its rapid results, simultaneous detection of rifampicin resistance (beneficial for multidrug-resistant TB (MDR-TB) screening), its high cost and infrastructure requirements restrict accessibility in resource-limited settings (Brown et al., 2021). Crucially, all sputum-dependent methods – including smear microscopy, culture, and GeneXpert – are limited by sample quality. Results are significantly compromised for patients who cannot produce sputum, produce scant sputum, or provide poor-quality samples (Oliveira da Silva et al., 2024; Kiflie et al., 2022). Non-sputum-based alternatives, such as urine lipoarabinomannan (LAM) detection, offer potential to overcome this limitation (Kasule et al., 2024).

Beyond diagnosis, monitoring treatment response remains a major challenge. Conventional methods, such as smear microscopy and Mycobacterial Culture, often fail to provide real-time insights into treatment efficacy, especially in patients who rapidly convert to negative early in therapy (Heyckendorf et al., 2022). This creates a critical gap in identifying patients at risk of treatment failure or relapse, highlighting the need for more reliable and accessible monitoring tools.

The urinary lipoarabinomannan (LAM) assay has emerged as a promising point-of-care test for TB diagnosis and monitoring. LAM, a component of the mycobacterial cell wall, is shed into urine during active infection, offering a non-invasive and easily accessible sample (Flores et al., 2021). Early-generation LAM tests, including the Abbott Determine TB LAM Ag (AlereLAM) and Fujifilm SILVAMP TB LAM Assay (FujiLAM), exhibited suboptimal sensitivity (Broger et al., 2020; Broger et al., 2023). AlereLAM demonstrated sensitivities of 42% in HIV-positive and 18% in HIV-negative population, while FujiLAM showed 70.7 and 53.2% (Broger et al., 2020) in these groups, respectively (Broger et al., 2023). This limited performance, particularly in HIV-negative individuals, restricted their utility primarily to HIV-positive patients with advanced immunosuppression (World Health Organization, 2019). These limitations stemmed from several factors: inadequate analytical sensitivity of lateral flow immunoassays, suboptimal antibody affinity in earlier test formulations (Li et al., 2024), cross-reactivity with nontuberculous mycobacteria (NTM) species (Calhoun et al., 2024) (particularly limiting its diagnostic utility for tuberculosis in people living with HIV) (Liu et al., 2022), and host immunity (Li et al., 2025).

Recent advancements in LAM technology, particularly the development of advanced chemiluminescence methods (AIMLAM), have addressed many of these limitations. By incorporating chemiluminescence detection and optimized urine processing protocols, AIMLAM achieves significantly enhanced sensitivity, exceeding 50–55% in HIV-negative populations—a performance comparable to GeneXpert MTB/RIF in some settings (Li et al., 2025; Gao et al., 2024). This technological breakthrough has expanded the potential utility of LAM testing beyond HIV-coinfected individuals to include the broader TB patient population.

While previous studies have demonstrated the diagnostic utility of AIMLAM (Li et al., 2025; Gao et al., 2024), its role in treatment monitoring remains largely underexplored. Existing evidence suggests that LAM levels may correlate with mycobacterial load and treatment response (Jones et al., 2022), but systematic comparisons with conventional microbiological methods are lacking.

Critical gaps persist regarding AIMLAM’s performance stratified by mycobacterial load, its comparative utility against standard microbiological methods during treatment, and its role as a predictive biomarker for treatment response in HIV-negative patients—who bear most of the global TB burden yet remain underserved by LAM diagnostics s (World Health Organization, 2024).

This study aims to address these critical gaps by: 1. evaluating the diagnostic performance of AIMLAM against a composite reference standard (CRS) compared to AFB, MGIT, and GeneXpert; 2. stratifying patients based on AFB and MGIT results to assess AIMLAM’s differential utility across disease severity spectra; and 3. establishing a predictive model for early treatment response using serial LAM measurements. Through integrating diagnostic accuracy, longitudinal monitoring, and predictive analytics, this research seeks to redefine the clinical utility of LAM testing in TB management.

The proactive identification of cases can reduce the transmission of tuberculosis, which necessitates the development of simpler and faster technologies to assist clinics in detecting more tuberculosis patients at an earlier stage, thereby curbing the spread of the disease (Ivanyi, 2014). This study provides a comprehensive evaluation of AIMLAM’s diagnostic and monitoring performance compared to conventional TB diagnostic methods (AFB smear, MGIT culture, and GeneXpert), while introducing a novel predictive model (PreLAM+ΔLAM) for early treatment response. The findings highlight AIMLAM’s unique dual utility in TB management—balancing diagnostic sensitivity with prolonged monitoring capability—and establish the clinical relevance of LAM antigen dynamics as a predictive biomarker for early treatment response.

AIMLAM demonstrated superior sensitivity (58.75%) to AFB smear (45.00%), aligning closely with MGIT culture (58.75%) and GeneXpert (61.25%), albeit with slightly lower specificity (96.67% vs. 100% for conventional methods). These findings are consistent with previously reported results demonstrating 51–55% sensitivity, significantly higher than AFB smear and comparable to culture-based sensitivity (Gao et al., 2024; Li et al., 2025; Huang et al., 2023). The present study found that in the Total, DP, and SP groups, GeneXpert exhibited marginally higher positive detection rates than AIMLAM before treatment (Pre) (61.25% vs. 58.75, 86.2% vs. 75.86, and 76% vs. 68%, respectively). However, in the DN group, AIMLAM showed a slightly higher rate than GeneXpert (30.77% vs. 11.54%). As a molecular diagnostic technique, GeneXpert typically achieves high sensitivity and specificity (Acharya et al., 2020). Its lower positive detection rate in the DN group may be attributed to the low bacterial load in samples (Hai et al., 2019) or poor sputum quality (Mulengwa et al., 2022), which could lead to false-negative results. This positions AIMLAM as a viable alternative in settings where rapid smear microscopy lacks sensitivity, and it holds advantages for populations with scanty sputum or difficulties in sputum collection, such as children and patients with extrapulmonary infections (Huang et al., 2023).

However, its most striking feature lies in its monitoring performance (Peter et al., 2016). While conventional methods (AFB, MGIT, GeneXpert) showed rapid declines in positivity rates and semi-quantitative levels within the first month of treatment—reflecting early bacterial load reduction—AIMLAM’s antigen concentrations decreased gradually, with significant reductions only at week 12 in the DP group. This delay may be attributed to: 1. Continuous LAM release from macrophage-engulfed bacilli undergoing lysosomal degradation (Ghazaei, 2018) and sterilizing granulomas where residual cell-wall fragments accumulate (Davuluri et al., 2023); 2. Potential undetected drug-resistant cases with suboptimal treatment response, despite rigorous exclusion of confirmed resistant tuberculosis through pre-enrollment (Italia et al., 2023). However, it should be emphasized that the clinical utility of LAM-based assays for monitoring treatment efficacy in EPTB remains unexplored, as no relevant studies have been reported to date.

GeneXpert exhibited the highest diagnostic sensitivity (61.25%), which is consistent with previous studies (Burger et al., 2022; Huang et al., 2023). However, its shorter conversion time (4.45 weeks vs. AIMLAM’s 7.85 weeks in the total cohort) underscores its limitation in long-term monitoring. Notably, a recent study reported that 80% of patients with AFB/MGIT/Xpert/MBLA quadruple-positive pulmonary TB remained positive by these methods at week 14 of treatment (Neumann et al., 2025), contrasting sharply with the 10.34% positivity rate observed in our DP group at week 12. This discrepancy may stem from differences in study populations: while the cited study enrolled patients uniformly positive across all four methods (AFB/MGIT/Xpert/MBLA), our DP group included only AFB/MGIT double-positive cases (n = 29 in both cohorts). The stricter inclusion criteria in the former—requiring positivity across sputum-dependent and molecular assays—likely selected for patients with higher baseline bacterial burdens, explaining the slower clearance rates (Neumann et al., 2025). Conversely, AIMLAM’s prolonged antigen detection window could fill a critical gap in evaluating end-of-treatment outcomes, particularly in high-burden DP patients, where its conversion time (11.59 weeks) far exceeded other methods.

Our study found that AIMLAM exhibited a specificity of 95.92% in non-TB controls, which was lower than that of AFB smear, MGIT culture, and Xpert MTB/RIF (all 100%). This observation aligns with findings from prior research (Huang et al., 2023). The slightly reduced specificity could potentially be explained by several factors: 1. Given the structural homology of LAM across mycobacterial species, the two false-positive cases might be attributed to either latent nontuberculous mycobacteria (NTM) contamination during sample processing—considering the abundance of NTM in the environment (Sharma and Upadhyay, 2020)—or subclinical NTM infection that escaped detection by routine diagnostic approaches. Huang et al. (2023) reported an AIMLAM positivity rate of 36.36% in confirmed NTM patients. Notably, values within the “gray zone” of the assay (i.e., ±15% of the cutoff) are inherently prone to measurement variability. One false-positive case, with a value of 0.50 U/mL relative to the 0.45 U/mL cutoff, falls precisely within this marginal range, which could further contribute to the observed result.

A critical advantage of AIMLAM lies in its non-sputum-dependent design. Unlike AFB smear, MGIT culture, and GeneXpert—which rely heavily on high-quality sputum samples and are thus limited in paucibacillary, pediatric, or extrapulmonary cases (Tendolkar et al., 2021), AIMLAM utilizes urine, a universally accessible specimen (Bulterys et al., 2019). This eliminates challenges associated with sputum collection (e.g., induction difficulties, contamination risks) and expands diagnostic access to underserved populations, including HIV-negative individuals with atypical presentations. Tuberculosis antibodies, including specific antibodies targeting LAM antigens, show promising potential in the diagnosis and therapeutic monitoring of tuberculosis (Bothamley et al., 1992). Ivanyi et al. found that antibody production is associated with the bacterial load of the disease, and antibodies have significantly higher sensitivity in detecting sputum smear-positive pulmonary diseases (Ivanyi, 2012). It has also been shown that antibody levels may increase during anti-tuberculosis treatment. Bothamley et al. observed that the antibody to the ML34 epitope of lipoarabinomannan consistently showed a single peak in antibody levels at 3 to 4 months (Bothamley, 2004). Analyzing the combined value of LAM antigens and antibodies in the diagnosis and therapeutic monitoring of tuberculosis may lead to new discoveries. However, due to the small number of tuberculosis antibody tests conducted in our hospital, they were not included in this study.

The combined PreLAM+ΔLAM model emerged as a novel and robust predictor of early LAM clearance, achieving AUC values >0.87 across all cohorts. Our Kaplan–Meier analysis revealed that low PreLAM+ΔLAM levels were strongly associated with accelerated LAM conversion across all patient cohorts. Notably, PreLAM+ΔLAM emerged as the most robust independent predictor in univariate and multivariate models for the total and double-positive cohorts. To our knowledge, no prior studies have specifically reported predictive models for short-term LAM conversion dynamics during anti-TB therapy. Baseline PreLAM levels likely reflect initial antigen burden and bacterial load, as supported by studies demonstrating correlations between urinary LAM concentrations and baseline sputum mycobacterial load in pulmonary TB (Paris et al., 2017). Meanwhile, ΔLAM (week 4 LAM change) captures early treatment response dynamics, with its magnitude potentially modulated by drug efficacy, host immune status (e.g., CD4 + T-cell function) (Correia-Neves et al., 2019), and metabolic clearance rates (Lawn et al., 2009; Wood et al., 2012). Clinically, this model enables risk stratification: patients with high PreLAM and stagnant ΔLAM may benefit from extended therapy or adjunctive interventions (e.g., immunomodulators), whereas those with low PreLAM and rapid ΔLAM decline could qualify for shortened regimens, reducing overtreatment.

The univariate Cox regression identified albumin and prealbumin as significant predictors in the total cohort, suggesting that nutritional status and systemic inflammation may play pivotal roles in determining treatment outcomes.

Some literature indicates that a too-low prealbumin level can assist in differentiating active tuberculosis from latent tuberculosis infection (Luo et al., 2021), and it has the potential to serve as an indicator for monitoring the therapeutic effect of tuberculosis treatment (Luo et al., 2013). Similarly, AFB smear grade, MGIT TPP, GeneXpert grade inversely correlated with LAM conversion rates, consistent with studies showing that high bacterial load predicts delayed culture conversion (Magombedze et al., 2021).

This study has several limitations. First, this study with an overall sample size of 130, stratifying into key subgroups (80 PTB vs. 50 non-TB; and finer divisions like 29 AFB+/MGIT+, 18 AFB-/MGIT+, 7 AFB+/MGIT-, 26 AFB-/MGIT-) drastically reduces the number of participants within each comparison group. This significantly diminishes statistical power, severely constraining our ability to conduct robust comparisons of methodological sensitivity and specificity across subgroups. Consequently, the risk of failing to detect statistically significant differences is heightened, and the reliability of estimates derived from smaller subgroups is reduced.

Second, the single-center design introduces potential biases related to local treatment protocols and patient population characteristics, which may limit the external validity and generalizability of the findings to broader or diverse healthcare settings and populations. Third, the 12-week observation period precludes evaluation of AIMLAM’s correlation with tuberculosis (TB) relapse or sustained cure (requiring ≥6 months of follow-up), and there is a lack of direct validation using clinical endpoints such as treatment success rates or mortality. Fourth, the current dataset has not yet integrated host immune biomarkers (e.g., cytokines, transcriptomics, CD4+/CD8 + ratios) to elucidate mechanisms underlying LAM dynamics, despite existing evidence showing LAM levels are closely associated with host immune status (e.g., delayed antigen clearance in HIV-coinfected individuals due to immunosuppression) (Wood et al., 2012; Flores et al., 2021). Future work will address these gaps through:

1. Large-scale, multicenter studies to enhance statistical power and generalizability; 2. Extended follow-up periods incorporating treatment completion and monitoring patients for ~1 year post-treatment success to capture relapse data; and 3. Integration of host immune parameters with the PreLAM+ΔLAM model. These steps aim to enable more precise TB management within clinical decision support systems.

This study positions urinary LAM as a pivotal dual-purpose biomarker in TB care, addressing unmet needs in both diagnosis and therapeutic monitoring. While demonstrating diagnostic performance comparable to established microbiological and molecular methods, LAM’s distinctive strength lies in its capacity to track treatment response beyond the early bactericidal phase. Unlike conventional sputum-dependent assays that rapidly decline in positivity, LAM’s prolonged detection window captures persistent antigen dynamics, reflecting residual bacillary activity—a critical advance for evaluating sustained therapeutic efficacy.

The integration of baseline antigen levels and early-treatment kinetic changes into a predictive model for early treatment response enables early identification of patients potentially requiring closer monitoring or therapy optimization during the initial treatment phase. This approach aligns with precision medicine paradigms, offering actionable insights for personalized intervention strategies early in the course of therapy.

By leveraging non-invasive urine sampling, LAM overcomes limitations of sputum-based methods, extending diagnostic and monitoring access to pediatric, paucibacillary, and extrapulmonary TB cases. Future validation through multicenter cohorts and correlation with long-term clinical endpoints will solidify its role in global TB control, advancing the WHO vision of biomarker-guided therapeutic optimization.