The tuberculosis (TB) epidemic is most prevalent among the elderly, with notification rates increasing progressively with age. Moreover, the mortality rate from tuberculosis remains higher in elderly patients. In China, national TB prevalence surveys have demonstrated that older people, especially older men, have the highest prevalence of TB of all age groups¹¹. Moreover, the diagnosis of TB in the elderly is more difficult due to nonspecific symptoms and less prominent clinical presentations. Consequently, elderly patients often require imaging and invasive procedures for definitive diagnosis. Awareness of risk factors, atypical manifestations, comorbidities, and appropriate evaluation is essential for early diagnosis of TB in elderly¹². This study highlights the diagnostic value of mNGS in adult and elderly PTB patients and explored age-related differences in lung microbial composition and clinical characteristics and imaging results. Understanding these differences may aid in the early diagnosis and treatment of TB, ultimately improving patient prognosis.

Several studies have explored the diagnostic value of mNGS for TB, yet no consensus has been reached regatding its detection capacity for MTB. In Zhang et al. study, mNGS demonstrated superior performance compared to GeneXpert¹³. However, other studies have found that mNGS does not outperform GeneXpert in detecting MTB^14,15. These discrepancies maybe influenced by factors such as sample types, whether patients undergoing anti-TB therapy, or difference in DNA extraction and host DNA depletion process within the mNGS protocol. In this study, a significant difference in the positive rate was observed between mNGS (100%) and GeneXpert (37.80%), which may largely be attributed to our preference for performing mNGS testing on patients with GeneXpert-negative results. T-spot is another commonly used TB diagnosis method which is susceptible to some factors, such as older patients, overweight and longer hospitalized¹⁶. In Liu et al. study, the positive rates of T-SPOT assay and mNGS were not statistically significant (P > 0.05)¹⁷. While Zhu et al.¹⁸ found increased performance of mNGS for TB identification compared to T-spot. Nevertheless, it is undeniable that mNGS exhibits higher sensitivity than AFS and culture for TB diagnosis^17,19. In this study, mNGS (100%) demonstrated superior performance in TB detection compared to T-spot (74.75%), GeneXpert (37.80%) and AFS (7.30%), and all the conventional methods showed a slight higher positive rate in the elderly group compared to the adult group (P > 0.05) (Fig. 2). Importantly, relying solely on conventional methods would have failed to diagnose all PTB cases in this study. These findings underscore the significance of employing multiple diagnostic methods, especially mNGS, for the accurate diagnosis of tuberculosis.

Rapid and precise detection of pathogenic microorganisms is crucial for the early diagnosis, appropriate treatment, and accurate prognosis of infectious diseases. Compared to commonly used clinical diagnostic methods such as smear microscopy and GeneXpert, which are cost-effective (costing only a few dollars per sample) and provide results within hours, mNGS has several drawbacks. These include a longer turnaround time (approximately 24 h) and higher costs (ranging from hundreds to thousands of dollars per sample). This is due to the complex processes involved in mNGS, including sample collection and pretreatment, microbial nucleic acid extraction, library preparation, high-throughput sequencing, bioinformatics analysis, and professional interpretation of results, as well as the use of expensive instruments. However, advancements in the standardization and automation of mNGS protocols are expected to significantly reduce costs, facilitating its widespread adoption in clinical settings for rapid and accurate diagnosis.

Disruptions in the host microbiome due to multiple intrinsic or extrinsic factors render people more susceptible to TB infection. These changes lead to reduced resistance to colonization by external pathogens or the loss of commensal bacteria, leading to pulmonary disease. Analysis of the lung microbiome in TB-positive and TB-negative cases revealed that alpha diversity was lower in the TB-positive group, while beta diversity also showed significant changes, and Mycobacterium and Anoxybacillus showed highly abundant in TB-positive cases, while TB-negative cases enriched with Prevotella, Alloprevotella, Veillonella, and Gemella²⁰. Several other studies have investigated the differences in lung microbiome composition between TB patients and healthy controls, and revealed that significantly changed microorganisms, including decrease of Streptococcus, Prevotella, and Fusobacterium, and increase of Lactobacillus, Acinetobacter, Mycobacterium, Staphylococcus, Neisseria, Veillonella, and Haemophilus in TB patients^21,22,23. However, studies investigating the composition and differences of lung microbiota in TB patients of different ages are still lacking. In our study, the within (alpha) and between (beta) sample diversity of the PTB microbiomes showed that microbial dysbiosis in the PTB patients is closely linked to different levels of age (Fig. 3A-C). The abundance of MTB was significantly higher in the elderly group than in the adult group (Fig. 3D). Several represent species were filtered out for the adult and elderly group, and some species of Neisseria, Veillonella, and Haemophilus showed increased abundance in the elderly group (with higher TB abundance)(Fig. 3E). Compared with their abundance changing between the TB and healthy patients, the results demonstrated that these microorganisms may play an important role in determining susceptibility tuberculosis infection.

Elderly pulmonary tuberculosis is characterized by a prolonged course, reduced immune function, and multiple complications, which can cause lesions in the lung interstitium and pulmonary blood vessels in addition to lesions in the lung parenchyma. The imaging manifestations of lung lesions are often multiple, polymorphic, and atypical, complicating disease diagnosis²⁴. Therefore, studying the imaging characteristics of elderly pulmonary tuberculosis is of great significance for the early detection, early diagnosis, and early treatment. In this study, we found the proportion of bilateral lung infection in elderly patients was significantly higher than that in adult patients. Furthermore, compared with the adult group patients, the frequency of bronchiectasis, emphysema, pleural effusion, lymph node enlargement or calcification, pulmonary bullae and pericardial effusion was significantly higher, but the frequency of infiltration and nodules was significantly lower in the elderly patients (Table 3). The incidence of bilateral lung diseases in elderly patients is often attributed to a decline in immune function with age, leading to decreased cell function and a greater susceptibility to infections and chronic lung conditions, making them more prone to developing bilateral lung diseases²⁵. Bronchiectasis is a common lung condition that can occur as a result of PTB²⁶. Emphysema can occur as a secondary effect of tuberculosis, causing destruction of lung tissue and resulting in reduced lung capacity²⁷. Pleural effusion is the second most common form of extrapulmonary tuberculosis. A previous study found that pleuritic chest pain (75%) and nonproductive cough (70%) were the most common symptoms in elderly patients with atypical cellular and biochemical features²⁸. While in our study, the similar results were not found. Moreover, atypical radiological features, such as middle or lower lobe (rather than upper lobe) infiltrates, mass-like lesions or nodules appearing more like cancers, extensive bronchopneumonia without cavitation or nonresolving infiltrates, are frequently misdiagnosed as pneumonia or lung cancer in the elderly²⁹. These study demonstrated that atypical imaging symptoms may also occur in elderly patients. At this time, we should pay attention to distinguishing tuberculosis from other lung diseases based on the patient’s clinical manifestations and etiological results.

Most positive correlations were found between the top 20 most abundant species, significantly different clinical characteristics, and imaging features (P < 0.05) between adult and elderly patients (Fig. 4). Among them, Streptococcus oralis showed positively correlation with age, and Prevotella veroralis and P. oris showed negative correlation with age. Previous study have revealed that Streptococcus were enriched in the PTB patients than healthy controls³⁰. Moreover, Streptococcus were also the most frequently detected bacterial phylotype in patients with different degrees of pulmonary emphysema³¹. Emphysema, which was more common in elderly patients than adults, were found positively correlated with S. oralis (Fig. 4). These demonstrated that elderly patients with emphysema and enriched S. oralis maybe more susceptible to TB infection. Tett et al.. found that the distribution of Prevotella spp. is also influenced by age³². Luo et al. revealed that a decrease of Prevotella in TB patients³³. These demonstrated that in TB pattients, the changing of Prevotella worth noting.

In this study, some patients have virus co-detected. Human herpesvirus 7 was the most common, followed by Epstein-Barr virus (EBV), Cytomegalovirus (CMV), human alphaherpesvirus 1 (HSV-1), human polyomavirus 5, human herpesvirus 6B, human adenovirus subtype C, human bocavirus, and HPgV type C. Herpes virus is commonly detected in patients with lung infection^34,35but the cases that cause disease are rare. Studies have found that CMV, like tuberculosis, can infect macrophages and dendritic cells. Under the influence of infection, these cells will reduce the production of IL-12, thereby disrupting the functional activity of Th1 response and aggravating the progression of tuberculosis³⁶. In addition, Th1 response may also be mediated by the action of IL-10, and IL-10 homologues are synthesized by CMV-infected cells³⁷. In adult tuberculosis patients, a study of adults from Russia found that the proportion of patients with pulmonary tuberculosis and herpes virus co-infection was higher than that of healthy patients, indicating a link between herpes virus infection and tuberculosis³⁸. This study did not use other detection methods to confirm whether the patients had viral infection, so the study of tuberculosis infection and viral infection is still worth conducting.

This study has several limitations, which are expected to be improved in future studies. On the one hand, we included relatively small sizes of cohorts, mainly because this is a single-center retrospective study. On the other hand, we included only samples with positive mNGS tuberculosis test results, which introduced a certain degree of bias and may have led to a biased assessment of the diagnostic performance of various methods. Finally, due to the long treatment period and difficulty in follow-up of tuberculosis patients, the use of antibiotics and treatment effects were not evaluated.