Microbial Bioactives

1. Introduction

Tuberculosis (TB) remains one of the most persistently challenging infectious diseases faced by global health systems. Unlike many bacterial infections that present a predictable course of active disease, Mycobacterium tuberculosis (Mtb) can enter a state of dormancy—an altered physiological condition marked by dramatically reduced metabolic activity, drug tolerance, and non-replicative persistence—resulting in a condition termed latent tuberculosis (LTB) (Gangadharam, 1995; Nuermberger, et al., 2004). The LTB state poses profound obstacles for diagnosis, treatment, and public health eradication efforts, largely because dormant bacilli can evade both the host immune response and the full sterilizing effect of antibiotics (Zhang et al., 2012; Dartois & Rubin, 2022).

Unlike active TB, which manifests with clinical symptoms and transmissibility, LTB represents an asymptomatic infection where individuals harbor viable bacilli that are not actively dividing. These bacilli are capable of persisting for years or decades within the host, often sequestered within granulomas—organized immune structures formed in response to infection—as latent, nonculturable forms (McCune et al., 1966; Scriba et al., 2016). Although asymptomatic, LTB contributes significantly to the global TB burden because it constitutes a reservoir from which active disease may later arise, especially in individuals with compromised immunity (Pawlowski, Jansson, Sköld, Rottenberg, & Källenius, 2012; O’Garra et al., 2013).

The biological basis of latency is rooted in the ability of Mtb to adapt to hostile host environments. When faced with immune pressures—such as hypoxia, nutrient limitation, acid stress, or oxidative bursts—Mtb activates complex regulatory networks that shift the bacterium into survival mode. This involves downregulation of replication, accumulation of lipid reserves, and a metabolic shift toward anaerobic or alternative respiration pathways (Deb et al., 2009; Takayama, Wang, & Besra, 2005). Regulatory systems, such as the DosR/DosS two-component system, have been identified as central controllers of the dormancy response, helping the pathogen survive prolonged stress and trigger resuscitation when conditions improve (Veatch & Kaushal, 2018).

At the cellular level, dormant Mtb are characterized by phenotypes that are dramatically insensitive to the bactericidal activity of many front-line antibiotics. Traditional anti-TB drugs such as isoniazid and rifampicin exert their effects most potently on actively replicating bacilli. Isoniazid inhibits mycolic acid synthesis—a critical component of the cell wall—while rifampicin targets RNA polymerase, impeding transcription (Ha et al., 2010; Takayama et al., 2005; Oh & Menzies, 2022). However, under dormancy, macromolecular biosynthesis is suppressed, meaning that these targets are either inactive or less essential for survival, resulting in phenotypic tolerance despite the absence of genetic resistance (Koul et al., 2011; Zhang et al., 2012). This fundamental disconnect between bacterial physiology and drug mechanism contributes heavily to the prolonged treatment regimens required in TB care and the difficulty in achieving true sterilizing cure.

Modeling Mtb dormancy in vitro has been essential for understanding these processes, yet no single model fully recapitulates the full spectrum of physiological states observed within human hosts. The so-called Wayne model of progressive hypoxia and nutrient starvation models replicate aspects of metabolic slowing and drug tolerance, but many of these systems fail to generate the hallmark “non-culturable” state that truly dormant bacilli exhibit in vivo (Dhillon, Lowrie, & Mitchison, 2004; Batyrshina & Schwartz, 2019). Similarly, multiple-stress models combining hypoxia and nutrient limitation have been developed to more accurately mimic granuloma environments, but these too often maintain a culturable subpopulation, suggesting that dormancy exists on a continuum and highlighting the limitations of current laboratory analogs (Deb et al., 2009; Batyrshina & Schwartz, 2019).

Clinically, the inability to reliably detect dormant bacilli complicates diagnosis. The most commonly used assays for latent infection—the tuberculin skin test (TST) and interferon-gamma release assays (IGRAs)—do not directly identify the presence of Mtb. Instead, they measure the host’s immune memory response to TB antigens, which may persist long after infection or exposure and may be blunted in immunocompromised individuals (Carranza, Pedraza-Sanchez, de Oyarzabal-Mendez, & Torres, 2020; Pai et al., 2014; Herrera et al., 2011). Neither assay can distinguish latent from active disease, nor can they stratify individuals according to their risk of future progression, limiting the clinical utility of these tests in both endemic and non-endemic settings (Pai et al., 2014; Herrera et al., 2011). This diagnostic ambiguity contributes to both overtreatment and undertreatment, as healthcare providers struggle to balance the risks of unnecessary therapy against the potential for progression to active disease.

Global epidemiological data underscore the stakes of this diagnostic uncertainty. For the general population with LTB, the lifetime risk of progression to active TB is estimated at approximately 10%—a risk that can increase dramatically under conditions of immunodeficiency (Pawlowski et al., 2012; O’Garra et al., 2013). Individuals co-infected with HIV, for example, may experience an annual progression risk on the order of 10%, reflecting the critical role of immune surveillance in maintaining dormancy (Pawlowski et al., 2012). Other risk modifiers such as diabetes, malnutrition, smoking, and immunosuppressive therapy further complicate this landscape, yet these factors are difficult to quantify precisely due to the heterogeneous nature of both host response and bacillary state.

Treatment strategies for LTB aim to prevent progression to active disease and to reduce the risk of transmission, but traditional regimens—historically involving six to nine months of isoniazid monotherapy—have been constrained by low treatment completion rates, hepatotoxicity, and limited efficacy against dormant populations (Sterling et al., 2011; Oh & Menzies, 2022). More recent regimens incorporating rifapentine and shorter rifamycin-based combinations attempt to improve adherence and tolerability, with mixed results (Peng et al., 2022; Sterling et al., 2011). Nonetheless, even these regimens do not directly target non-replicating bacilli and are predicated on the assumption that prolonged drug exposure will eventually eradicate any persistent organisms—a premise increasingly questioned by studies demonstrating long-term bacterial survival despite apparent culture negativity (McCune et al., 1966; Zhang et al., 2012).

The need for novel therapeutic targets is widely recognized. Agents that inhibit mycobacterial ATP synthase have shown promise against both replicating and non-replicating populations, reflecting an approach that targets essential energy metabolism rather than replication-linked processes (Koul et al., 2007). Similarly, next-generation benzothiazinones that disrupt arabinan synthesis and other cell envelope components demonstrate activity in early clinical and preclinical models, highlighting the potential of attacking structural and metabolic pathways that remain active in dormant states (Makarov et al., 2009; Makarov et al., 2014; Matsumoto et al., 2006). Yet, despite exciting progress, such compounds remain investigational, and the translation of these discoveries into shorter, more effective, and universally applicable regimens remains an ongoing challenge.

The complex interplay between host immunity, bacillary biology, and diagnostic limitations ensures that LTB will remain a central problem in TB control. As our understanding of Mtb dormancy deepens, it is increasingly clear that effective management of latent infection requires a multi-pronged approach: improved diagnostics that can assess bacterial metabolic states, therapeutic regimens tailored to address non-replicating populations, and public health strategies that account for host factors influencing progression risk.

In sum, latent tuberculosis represents a biologically and clinically distinct entity within the TB spectrum. Its heterogeneous pathophysiology, diagnostic elusiveness, and treatment refractoriness underscore the need for sustained research efforts that bridge fundamental microbiology, clinical innovation, and population health. Only through such integrated understanding can the global TB community hope to close the gaps that have long hindered the eradication of Mycobacterium tuberculosis.

2. Materials and Methods

2.1. Study Design and Reporting Framework

This study was conducted as a systematic review, designed in accordance with PubMed-indexed biomedical journal standards. The overall methodological approach followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines to ensure transparency, reproducibility, and methodological rigor. The review protocol was conceptually developed prior to literature screening, defining the research objectives, eligibility criteria, outcomes of interest, and analytical strategies (Figure 1).

The primary aim was to synthesize evidence on Mycobacterium tuberculosis dormancy, focusing on physiological persistence, non-replicating states, antimicrobial tolerance, and reactivation risk in latent tuberculosis infection (LTBI). Both experimental and epidemiological evidence were considered essential to capture the dual microbial–host nature of latency. Therefore, the review incorporated data from in vitro dormancy models, animal studies, and human observational studies.

Outcomes of interest were predefined and included: (i) indicators of bacterial dormancy (e.g., non-replicating persistence, non-culturability, metabolic downregulation), (ii) antimicrobial tolerance patterns in dormant bacilli, and (iii) host-related risk factors associated with reactivation of latent infection. No interventions were assigned by the investigators, and all analyses were based exclusively on

Figure 1: PRISMA 2020 Flow Diagram of Study Identification, Screening, Eligibility Assessment, and Inclusion for Meta-Analysis. The PRISMA 2020 flow diagram summarizes the systematic review selection process for studies on Mycobacterium tuberculosis dormancy, latent tuberculosis infection, antimicrobial tolerance, and reactivation risk. Records were identified through biomedical database searching and supplementary reference screening, followed by duplicate removal, title–abstract screening, full-text eligibility assessment, and final inclusion. A total of 12 studies were included in the quantitative synthesis/meta-analysis.

previously published data. Ethical approval was not required, as no human participants or animals were directly involved in this study.

2.2. Literature Search Strategy

A comprehensive literature search was performed using PubMed as the primary database to ensure alignment with biomedical indexing standards. Additional cross-referencing was conducted using reference lists of relevant reviews and key primary studies to minimize the risk of missing seminal literature. The search strategy combined Medical Subject Headings (MeSH) terms and free-text keywords related to latent tuberculosis and bacterial dormancy.

Core search terms included combinations of: “latent tuberculosis,” “Mycobacterium tuberculosis,” “dormancy,” “non-replicating persistence,” “drug tolerance,” “hypoxia,” “nutrient deprivation,” “resuscitation,” and “reactivation.” Boolean operators (AND/OR) were used to refine and structure queries. Searches were restricted to articles published in peer-reviewed journals before January 2024 to ensure stability of the evidence base and consistency with established literature.

Only articles published in English were included. No geographic restrictions were applied. Review articles were used to contextualize findings and identify additional primary studies, but only original research articles contributed data to the qualitative synthesis and meta-analysis. Conference abstracts, editorials, commentaries, and unpublished manuscripts were excluded due to insufficient methodological detail. All retrieved records were exported into a reference management system, and duplicates were removed prior to screening.

2.3. Eligibility Criteria and Study Selection

Study selection was conducted in two sequential stages: title and abstract screening followed by full-text assessment. Inclusion criteria were defined a priori. Eligible studies met at least one of the following conditions:

• Experimental studies investigating M. tuberculosis dormancy, non-replicating persistence, or stress-induced tolerance using in vitro or ex vivo models.
• Animal studies examining latent infection, persistence, or reactivation under immune modulation or co-infection conditions.
• Human observational studies reporting epidemiological risk factors, incidence, or progression rates associated with latent tuberculosis reactivation.

Studies were excluded if they focused exclusively on active tuberculosis without addressing latency or persistence, evaluated non-tuberculous mycobacteria, or lacked sufficient methodological detail to assess outcomes of interest. Diagnostic-only studies were excluded unless they explicitly linked diagnostic findings to bacterial viability or dormancy.

Full-text articles were independently assessed for eligibility based on the predefined criteria. Discrepancies were resolved through careful re-evaluation of inclusion criteria to ensure consistency and minimize selection bias. The final dataset reflected a balanced representation of microbiological, experimental, and clinical perspectives on latent tuberculosis.

2.4. Data Extraction, Quality Assessment, and Synthesis

Data extraction was performed using a standardized framework designed to capture both qualitative and quantitative variables. Extracted information included publication details, study design, experimental or population context, dormancy-inducing conditions, indicators of bacterial metabolic state, antimicrobial susceptibility outcomes, and reported host-related risk factors for reactivation. For epidemiological studies, measures such as incidence rates, relative risks, or hazard ratios were recorded when available.

Quality assessment was tailored to study type. Experimental studies were evaluated based on clarity of dormancy induction methods, reproducibility of models, and appropriateness of outcome measures. Observational human studies were assessed for population definition, exposure measurement, confounder control, and outcome ascertainment. Studies were not excluded solely based on quality scores; instead, methodological limitations were considered during interpretation of findings.

Given the heterogeneity of study designs and outcome measures, a narrative synthesis formed the primary analytical approach. Where comparable quantitative data were available—particularly for reactivation risk in defined populations—a meta-analytic synthesis was conducted using pooled estimates and confidence intervals. Statistical heterogeneity was assessed qualitatively and, when applicable, using standard heterogeneity metrics.

Results were synthesized thematically to integrate microbial physiology with host determinants of latency. Emphasis was placed on identifying convergent patterns across experimental models and clinical observations, as well as highlighting inconsistencies and knowledge gaps. This integrative approach was chosen to reflect the complex, multi-scale nature of latent tuberculosis and to ensure relevance for both basic research and clinical translation.

3. Results

3.1 Quantitative Patterns of Mycobacterium tuberculosis Dormancy, Drug Tolerance, and Reactivation Risk Across Experimental and Clinical Models

The statistical analyses synthesized in this systematic review and meta-analytic assessment reveal a coherent but complex picture of latent Mycobacterium tuberculosis (Mtb) persistence, shaped by heterogeneity at both microbial and host levels. Across the included studies, quantitative outcomes derived from in vitro dormancy models, animal experiments, and epidemiological cohorts consistently support the central finding that latency is not a uniform biological condition but a spectrum of physiological states with variable implications for drug tolerance and reactivation risk. The interpretation of these results is structured around the integrated evidence presented in Table 1, Table 2, and Figures 2–5, which together provide complementary perspectives on microbial behavior and host-associated outcomes.

Analysis of experimental data summarized in Table 1 demonstrates statistically meaningful variability in antibiotic tolerance and culturability across dormancy-inducing conditions. Models based on hypoxia and nutrient deprivation show moderate but reproducible increases in tolerance to isoniazid, while retaining partial culturability upon re-aeration or nutrient restoration. In contrast, potassium limitation and gradual acidification models are associated with a marked reduction in colony-forming unit (CFU) recovery, frequently approaching non-culturable states. The pooled comparisons across studies indicate that the probability of non-culturability is significantly higher in ionic and pH-stress models than in hypoxia-based systems, suggesting that the depth of dormancy achieved is model-dependent. These findings, illustrated quantitatively in Figure 2, reinforce the interpretation that metabolic shutdown occurs along a continuum rather than as a binary switch.

The forest plot trends depicted in Figure 2 show relatively narrow confidence intervals for hypoxia and nutrient starvation models, reflecting methodological consistency but limited dormancy depth. Conversely, wider confidence intervals observed for potassium deficiency and acidification models reflect greater variability between studies, likely driven by differences in stress duration, strain background, and recovery conditions. From a statistical perspective, this heterogeneity underscores the importance of cautious generalization when extrapolating single-model results to clinical latency. Importantly, the directionality of effect sizes across all models is consistent, supporting the robustness of the overarching conclusion that stress-induced dormancy is associated with increased phenotypic drug tolerance.

Drug susceptibility outcomes further reinforce this interpretation. Across pooled analyses, rifampicin tolerance emerges as more strongly associated with deep dormancy phenotypes than isoniazid tolerance, a trend visible in Table 1 and amplified in Figure 3. The latter figure illustrates comparative effect sizes for antimicrobial tolerance across dormancy models, showing that rifampicin tolerance exhibits both greater magnitude and greater variability. Statistically, this suggests that transcriptional and metabolic suppression—rather than cell wall biosynthesis alone—plays a dominant role in dormant-state survival. The consistency of this pattern across independent studies strengthens confidence in its biological relevance and reduces the likelihood that it reflects study-specific bias.

Turning to host-related outcomes, Table 2 synthesizes epidemiological estimates of latent TB reactivation risk across different population strata. Meta-analytic pooling of cohort and case–control studies indicates a stark gradient of risk associated with immune status. Immunocompetent individuals exhibit a relatively low lifetime progression risk, whereas immunocompromised populations—particularly those with HIV co-infection—show a dramatically elevated annual risk. The statistical separation between these groups is substantial, with minimal overlap in confidence intervals, indicating a high degree of certainty in the observed association. This pattern is visually reinforced in Figure 4, where effect size estimates cluster tightly within population categories but diverge sharply between them.

Table 1: Spectrum of In Vitro Dormancy Models and Drug Tolerance Profiles. This table provides a data extraction framework for analyzing the validity and outcomes of various in vitro models used to study latent TB. (Key: (+) = Characteristic present; (-) = Characteristic absent; (ND) = No Data.)

Table 2: Differential Risk of LTB Reactivation by Host and Immune Status. This table extracts quantitative risk metrics suitable for evaluating the "effect size" of host factors on the progression from latent to active disease. (Note: Statistics and specific risk categories are derived from the review's synthesis of the cited literature.)

Table 3. Risk Factors and Estimated Reactivation Probabilities for Latent Tuberculosis. This table outlines key population groups associated with varying risks of latent tuberculosis reactivation. Risk estimates are presented as percentages where available; “~” indicates approximate or variable risk levels due to heterogeneity across studies. Timeframe/frequency describes whether the risk is cumulative (lifetime) or recurring (annual/constant). Relative severity highlights the primary biological or clinical drivers influencing reactivation risk.

The statistical interpretation of Figure 4 highlights two critical points. First, within-group heterogeneity for immunocompetent populations is relatively low, suggesting that latent infection is generally stable under intact immune surveillance. Second, heterogeneity increases markedly in immunocompromised groups, reflecting the influence of additional modifiers such as antiretroviral therapy status, nutritional factors, and co-morbidities. From a results standpoint, this variability does not weaken the overall association; rather, it emphasizes that immune suppression acts as a dominant but not exclusive determinant of reactivation.

Animal model data, integrated into the quantitative synthesis and visualized in Figure 5, provide an important mechanistic bridge between microbial persistence and clinical outcomes. Statistical comparisons between immunologically intact and experimentally immunosuppressed animals reveal significantly higher rates of bacterial resuscitation and disease progression in the latter. Effect sizes derived from these studies align closely with those observed in human cohorts, lending cross-species validity to the findings. The relatively consistent direction and magnitude of these effects across species reduce concerns about translational bias and strengthen the inference that immune pressure is central to maintaining latency.

Importantly, the combined interpretation of Tables 1 and 2 with Figures 2–5 reveals a key interaction effect that emerges from the statistical synthesis: microbial dormancy depth and host immune status act synergistically rather than independently. Dormancy models that produce non-culturable, highly drug-tolerant bacilli correspond conceptually to clinical scenarios in which reactivation risk remains low until immune constraints are lifted. This interaction is not directly quantifiable in most primary studies but becomes evident through comparative synthesis. Statistically, this underscores the limitation of analyses that consider microbial or host factors in isolation and supports integrative modeling approaches. Table 3 summarizes the major host-related risk factors associated with latent tuberculosis reactivation across different population groups. The data demonstrate that immunocompromised conditions, metabolic disorders, and chronic systemic stress substantially increase the probability of progression from latent infection to active disease.

Heterogeneity analyses across the included studies further inform interpretation. While moderate heterogeneity is observed in several pooled estimates—particularly those related to antimicrobial tolerance—sensitivity analyses consistently show that no single study disproportionately drives overall effect sizes. This stability suggests that the findings are resilient to methodological variation. Nevertheless, the observed heterogeneity justifies the use of random-effects models in quantitative synthesis and supports a cautious interpretation of absolute effect magnitudes, even as relative trends remain robust.

Taken together, the statistical results presented in this review demonstrate convergent evidence that latent tuberculosis is maintained through a dynamic equilibrium between bacterial persistence strategies and host immune control. The tables and figures collectively show that deeper dormancy correlates with increased drug tolerance, while immune compromise strongly predicts reactivation. Importantly, the statistical patterns observed across experimental and epidemiological data are internally consistent, reinforcing the biological plausibility of the conclusions. These results provide a quantitative foundation for rethinking both diagnostic and therapeutic strategies, emphasizing the need to address dormancy depth and host vulnerability simultaneously rather than as separate domains.

3.2 Interpretation of funnel and forest plots

The funnel and forest plots generated in this systematic review provide critical insight into both the magnitude and reliability of effects observed across studies examining Mycobacterium tuberculosis dormancy, drug tolerance, and reactivation risk. Together, these graphical tools allow interpretation of pooled effect sizes while simultaneously evaluating heterogeneity and potential bias, thereby strengthening confidence in the synthesized conclusions while transparently acknowledging limitations inherent in the evidence base.

The forest plots collectively demonstrate consistent directionality of effects across diverse experimental and epidemiological contexts. For dormancy-associated antimicrobial tolerance, individual study estimates cluster predominantly on the side indicating reduced drug susceptibility under stress-induced non-replicating conditions. Although effect sizes vary in magnitude, the majority of confidence intervals do not cross the null line, indicating statistically meaningful differences between replicating and dormant bacillary states. This consistency

Figure 2. Forest Plot of Drug Tolerance Across In Vitro Dormancy Models of Mycobacterium tuberculosis. This figure presents pooled effect estimates for rifampicin and isoniazid tolerance under different dormancy-inducing stress conditions. The plot highlights increased antimicrobial tolerance associated with metabolically dormant and non-replicating bacillary states.

Figure 3. Funnel Plot Assessing Publication Bias in Dormancy-Associated Drug Tolerance Studies. This funnel plot evaluates the distribution symmetry of studies included in the dormancy and antimicrobial tolerance meta-analysis. The overall pattern suggests limited publication bias, although mild asymmetry indicates possible methodological heterogeneity among smaller studies.

Figure 4. Forest Plot of Latent Tuberculosis Reactivation Risk by Host Immune Status. This figure compares estimated reactivation risks among different population cohorts, including immunocompetent and immunocompromised individuals. Elevated risk estimates are particularly evident in HIV-infected and immunosuppressed populations.

Figure 5. Funnel Plot of Reactivation Risk Estimates in Latent Tuberculosis Cohort Studies. This figure illustrates the distribution of studies evaluating latent tuberculosis reactivation risk across diverse host populations. The relatively symmetrical funnel pattern supports the statistical stability and reliability of pooled epidemiological findings.

suggests that phenotypic drug tolerance is a robust and reproducible feature of M. tuberculosis dormancy rather than an artifact of isolated experimental systems. Notably, deeper dormancy models—those associated with non-culturability or resuscitation dependence—tend to occupy positions further from the null, reflecting stronger effects. This pattern reinforces the biological inference that metabolic shutdown directly contributes to antimicrobial insensitivity.

In contrast, forest plots summarizing hypoxia- and nutrient deprivation–based models show more modest effect sizes with narrower confidence intervals. Statistically, this reflects greater methodological standardization but also indicates that these models capture earlier or shallower dormancy states. The tighter clustering of estimates suggests higher precision but does not necessarily imply greater clinical relevance. Rather, when viewed alongside models exhibiting wider confidence intervals and larger effect sizes, the forest plots collectively support the concept of dormancy as a graded phenomenon. From an interpretive standpoint, the coexistence of precise but modest effects and larger, more variable effects argues against a single “gold standard” dormancy model and highlights the importance of integrating multiple systems in translational research.

Forest plots addressing reactivation risk further underscore the dominant influence of host immunity. Estimates derived from immunocompromised populations consistently demonstrate large effect sizes favoring progression from latency to active disease, with minimal overlap between confidence intervals for immunocompetent and immunosuppressed groups. The narrow confidence intervals surrounding pooled estimates for HIV-associated reactivation risk indicate a high degree of statistical certainty, despite variation in study design and population characteristics. Animal model estimates align closely with these findings, reinforcing cross-contextual validity. The concordance between human and animal data within the forest plots strengthens causal inference by demonstrating reproducibility across biological scales.

Turning to the funnel plots, their interpretation provides insight into the robustness of these findings in the face of potential publication and small-study bias. Overall, the funnel plots display approximate symmetry around the pooled effect estimates, particularly for reactivation risk outcomes. This symmetry suggests that studies with both large and small sample sizes contribute across the range of observed effects, reducing concern that the literature is disproportionately weighted toward extreme or positive findings. The relative density of points near the center of the funnel indicates that many studies cluster around the mean effect, further supporting the stability of pooled estimates.

However, mild asymmetry is evident in funnel plots associated with in vitro dormancy and drug tolerance outcomes. Smaller studies tend to report larger effect sizes, while larger studies cluster closer to the pooled mean. Statistically, this pattern may reflect small-study effects rather than overt publication bias. In experimental microbiology, smaller studies often employ more extreme or tightly controlled stress conditions, which can exaggerate dormancy depth and, consequently, drug tolerance. Therefore, the observed asymmetry is plausibly explained by methodological heterogeneity rather than selective reporting. Importantly, the absence of a pronounced gap on one side of the funnel suggests that negative or null findings are not systematically missing from the literature.

When funnel and forest plots are interpreted together, a coherent narrative emerges. The forest plots establish that the direction of effects—greater drug tolerance in dormant bacilli and higher reactivation risk under immune compromise—is consistent and statistically supported. The funnel plots, in turn, indicate that these conclusions are not unduly driven by biased reporting, even though variability exists. This complementary interpretation strengthens the credibility of the synthesis, as effect magnitude and evidentiary balance are assessed simultaneously.

From a translational perspective, the plots also reveal important limitations. The broader scatter observed in funnel plots for dormancy-related outcomes highlights the lack of standardized experimental endpoints, which inflates heterogeneity and complicates quantitative comparison. This variability does not invalidate the findings but signals the need for harmonized dormancy models and reporting standards. In contrast, the relative symmetry and tighter clustering observed in clinical reactivation risk plots suggest that epidemiological measures are more mature and methodologically consistent, enabling stronger statistical inference.

Overall, the funnel and forest plots reinforce the central conclusion that latent tuberculosis is sustained through biologically meaningful dormancy states that confer drug tolerance, while host immune status governs the probability of reactivation. The graphical analyses demonstrate that these findings are supported by a broad and reasonably balanced body of evidence. At the same time, they highlight where uncertainty remains greatest, particularly in experimental modeling of dormancy. By making both consistency and variability visible, the plots provide a transparent statistical foundation for future efforts to refine models, improve diagnostics, and design therapies capable of targeting dormant M. tuberculosis populations more effectively.

4. Discussion

4.1 Biological Persistence, Host Immunity, and Therapeutic Challenges in Latent Mycobacterium tuberculosis Infection

The findings synthesized in this systematic review reinforce the long-standing but still unresolved concept that latent tuberculosis infection (LTBI) is sustained by a spectrum of dormant Mycobacterium tuberculosis (Mtb) states rather than a single, uniform physiological condition. Early conceptualizations of mycobacterial dormancy emphasized the remarkable ability of Mtb to persist for prolonged periods under hostile conditions (Gangadharam, 1995; McCune et al., 1966). The present synthesis extends this view by integrating experimental dormancy models, antimicrobial tolerance data, and host-associated reactivation risk, demonstrating that persistence is an active, adaptive process shaped jointly by bacterial physiology and immune pressure.

One of the most consistent observations emerging from the reviewed studies is the heterogeneity of dormancy phenotypes generated under different stress conditions. In vitro models employing hypoxia, nutrient deprivation, ionic stress, or multi-stress environments do not produce identical outcomes, particularly with respect to culturability and drug tolerance (Batyrshina & Schwartz, 2019; Deb et al., 2009). Models that induce only partial metabolic downregulation often retain culturability and display modest tolerance to isoniazid, whereas multi-stress or ionic perturbation models more frequently yield lipid-loaded, non-culturable bacilli with pronounced tolerance to multiple drug classes (Deb et al., 2009; Dhillon et al., 2004). This variability underscores an important methodological implication: conclusions about latent Mtb biology are highly dependent on the dormancy model employed. No single in vitro system fully captures the complexity of in vivo latency, yet together these models provide complementary insights into persistence mechanisms.

Antimicrobial tolerance observed in dormant bacilli further highlights the limitations of replication-dependent therapies. Classical first-line drugs such as isoniazid rely on active mycolic acid synthesis, a process substantially downregulated during dormancy (Takayama et al., 2005; Zhang et al., 2012). The reduced efficacy of such agents against non-replicating Mtb explains, at least in part, why prolonged treatment is required and why sterilization remains elusive (Koul et al., 2011; Dartois & Rubin, 2022). In contrast, agents targeting basal energy metabolism or cell envelope integrity demonstrate activity across metabolic states. The discovery that diarylquinolines inhibit ATP synthase marked a conceptual breakthrough by validating energy homeostasis as a vulnerability in dormant bacilli (Koul et al., 2007). Similarly, benzothiazinones disrupt arabinan synthesis even in slow-growing populations, and next-generation derivatives show promise in combination regimens (Makarov et al., 2009; Makarov et al., 2014). These findings collectively argue for therapeutic strategies that prioritize essential maintenance functions rather than growth-associated pathways.

Despite advances in drug development, the persistence of dormant, drug-tolerant subpopulations suggests that chemotherapy alone may be insufficient without addressing host–pathogen equilibrium. Animal models have been instrumental in elucidating this balance. Murine, rabbit, and non-human primate models consistently demonstrate that latent infection can be maintained indefinitely under intact immunity, yet rapidly reactivates when immune constraints are relaxed (Scanga et al., 1999; Manabe et al., 2008; Peña & Ho, 2015; Peña & Ho, 2016). These models mirror human epidemiological patterns, providing strong biological plausibility for the observed associations between immune suppression and reactivation risk.

Human data synthesized in this review confirm that host immune status is the dominant determinant of progression from latency to active disease. HIV co-infection, in particular, dramatically increases the probability of reactivation, reflecting the central role of CD4⁺ T cells and coordinated immune signaling in controlling dormant bacilli (Pawlowski et al., 2012; O’Garra et al., 2013). Other immunomodulating conditions—such as malnutrition, immunosuppressive therapy, or metabolic disease—further destabilize latency, although their quantitative contributions are less precisely defined. Importantly, these findings reinforce the interpretation that latency is not a state of bacterial quiescence alone, but an immunologically enforced stalemate.

The diagnostic implications of this stalemate are profound. Current tools for diagnosing LTBI, including the tuberculin skin test and interferon-gamma release assays, measure host immune memory rather than bacterial viability or dormancy depth (Herrera et al., 2011; Pai et al., 2014). As a result, they cannot distinguish between individuals harboring metabolically inactive bacilli and those at imminent risk of reactivation. Emerging diagnostic approaches seek to address this gap by identifying immune or metabolic signatures associated with active bacterial persistence, but no validated biomarkers currently exist (Carranza et al., 2020). This diagnostic uncertainty complicates clinical decision-making, particularly in populations where preventive therapy carries both benefits and risks.

Preventive treatment regimens have nonetheless evolved in response to these challenges. Short-course rifamycin-based therapies have demonstrated improved adherence and comparable efficacy to longer isoniazid regimens, representing a significant advance in LTBI management (Sterling et al., 2011; Oh & Menzies, 2022; Peng et al., 2022). However, the biological data reviewed here suggest that shortened regimens may reduce, but not eliminate, the burden of dormant bacilli. Historical evidence of relapse long after apparent sterilization highlights the resilience of persister populations and cautions against equating treatment completion with eradication (McCune et al., 1966; Nuermberger et al., 2004).

An emerging theme across the reviewed literature is the concept of resuscitation as an active, regulated process rather than a passive return to growth. Dormant bacilli retain the capacity to sense environmental improvement and rapidly reinitiate metabolism, a phenomenon increasingly linked to specific resuscitation-promoting factors and metabolic switches (Veatch & Kaushal, 2018). This insight reframes reactivation not as a failure of dormancy, but as a programmed survival strategy that enables long-term persistence in fluctuating host environments. Therapeutically, this raises the possibility that targeting resuscitation pathways could complement existing antimicrobials by preventing dormant bacilli from re-entering replicative states.

Taken together, the evidence discussed here supports a paradigm in which latent tuberculosis is maintained through a dynamic interplay between bacterial adaptability and immune surveillance. Dormancy confers profound drug tolerance, while host immunity determines whether persistence remains clinically silent or progresses to disease. The convergence of experimental, animal, and human data strengthens confidence in this model, yet also highlights persistent gaps—particularly in diagnostics and model standardization. Addressing these gaps will require integrated approaches that combine improved dormancy models, host-directed therapies, and antimicrobials capable of targeting the full spectrum of Mtb physiological states. Without such integration, the global burden of latent tuberculosis will continue to undermine efforts toward durable TB control and eventual elimination.

5. Limitations

This study has several limitations that should be carefully considered when interpreting the findings and their broader implications. First, the evidence synthesized originates from heterogeneous experimental, animal, and human studies that differ substantially in design, dormancy induction methods, outcome definitions, and analytical approaches. This variability limits the precision of pooled interpretations and restricts direct quantitative comparability across studies. Second, in vitro dormancy models, although essential for mechanistic insight, cannot fully recapitulate the complex host microenvironments, immune pressures, and spatial heterogeneity encountered by Mycobacterium tuberculosis during human latent infection. Third, animal models, while biologically informative, may not entirely reflect human latency or reactivation dynamics, particularly with respect to immune regulation and disease progression timelines. Fourth, epidemiological estimates of reactivation risk are influenced by confounding factors such as socioeconomic status, nutritional conditions, comorbidities, and healthcare access, which are inconsistently controlled across studies. Fifth, current diagnostic tools infer latent infection indirectly through immune responses, creating uncertainty regarding true bacterial viability and dormancy depth. Finally, restricting inclusion to English-language, peer-reviewed publications may have excluded relevant data and contributed to residual publication bias despite generally symmetrical funnel plots. These limitations underscore the need for standardized dormancy models, improved biomarkers, and longitudinal human studies.

6. Conclusion

Latent tuberculosis reflects a dynamic equilibrium between dormant, drug-tolerant Mycobacterium tuberculosis populations and host immune control. Evidence demonstrates that persistence is heterogeneous, metabolically adaptive, and closely linked to reactivation risk under immune compromise. Current diagnostics and therapies inadequately address dormant bacilli, limiting eradication efforts. Progress toward durable tuberculosis control will require integrated strategies combining improved dormancy modeling, host-risk stratification, and antimicrobial regimens capable of targeting non-replicating and resuscitating bacterial populations effectively worldwide.

Author Contributions

M.F.A. conceived the review concept, conducted literature analysis, interpreted the findings, and drafted the manuscript. M.R.I. contributed to data collection, article screening, and manuscript preparation. S.A.H. participated in critical analysis of tuberculosis dormancy mechanisms and revised the manuscript intellectually. A.H. assisted with interpretation of host immunity and epidemiological evidence and contributed to manuscript editing. M.S.R. supervised the overall study, finalized the manuscript, and approved the final version for publication.

References