Researchers found a three-fold increase in the incidence of detecting pulmonary tuberculosis (PTB) in patients hospitalised with COVID-19 pneumonia

Researchers from the Department of Epidemiology, Faculty of Medicine, Prince of Songkhla University-Thailand have in a new study validated that there is an increased incidence of detecting pulmonary tuberculosis (PTB) in patients hospitalized with COVID-19 pneumonia compared with that in the general population.

The study findings were published in the peer reviewed journal: eClinical Medicine (Lancet Discovery Science).
https://www.thelancet.com/journals/eclinm/article/PIIS2589-5370(23)00002-0/fulltext

A previous South African hospital-based study reported a 3-fold increase in the incidence of pulmonary tuberculosis (PTB) in those previously inflicted with COVID-19 pneumonia compared with that in the general population.
https://pubmed.ncbi.nlm.nih.gov/35725387/

The cumulative incidence of PTB among patients with SARS-CoV-2 infection was steep at ten-fold of the population control in the early period (within 30 days after infection), and it declined to six-fold thereafter. In the second period, patients with symptomatic COVID-19 (both with pneumonia and without pneumonia) demonstrated an increased hazard of detectable active PTB.

Also, the negative control had a slightly increased hazard in the early period. These would be patients who had PTB without SARS-CoV-2 co-infection. However, the degree of increase in the hazard subsided later in the follow-up. Only COVID-19 pneumonia was persistently associated with the hazard of detectable active PTB.

With accurate timing of the events of SARS-CoV-2 infection and the ascertainment of active PTB, our data removed the doubt on the chronological sequence found in the previous study.21 The differences in the hazard for detectable active PTB among groups with positive RT-PCR results for the SARS-CoV-2 virus may be explained by two main arguments: biological effects of SARS-CoV-2 infection and the difference in the chance of being investigated for PTB, called Berkson’s bias.23

The correlation between TB susceptibility and SARS-CoV-2 infection could be potentially ruled out as the underlying diseases were already included in the multivariable model.

Several studies have demonstrated that T cells, especially CD8+, play a significant role in the defence against TB.12, 13, 14, 15, 16 Two studies in China asserted that there is a significantly decreased number of CD8+ cells at the time of SARS-CoV-2 infection.48,49 The depletion of CD4+ and CD8+ in the early period of the SARS-CoV-2 infection has a dose–response relationship with the severity of COVID-19.48,49

Thus, the severity of COVID-19 could be a surrogate variable for the T-cell count affecting clinical presentation. Additionally, a study in China reported that 73.6% of patients with pneumonia could recover their T-cell numbers within approximately 30 days, whereas the remaining 26.4% had prolonged T-cell depletion.50 Decreased cellular immune response due to T-cell depletion could mediate the increased risk of getting active TB.

These findings could explain why patients with COVID-19 pneumonia may be at increased risk of subsequent PTB even if their COVID-19 symptoms are cured.

On the other hand, an in vitro study reported that patients with COVID-TB co-infection have a decreased response to SARS-CoV-2 and an intact Mycobacterium tuberculosis (Mtb) response. Moreover, COVID-19 patients with latent TB infection (LTBI) seem to have the ability to react to both SARS-CoV-2 and Mtb antigens.51

Thus, the increased hazard of detectable active PTB in COVID-19 pneumonia was less likely to be from reactivation of LTBI due to the COVID-LTBI interaction but more likely to be from the weakened immune system as the aforementioned. However, as we could not identify individuals’ spectrum of TB infection before SARS-CoV-2 infection, this hypothesis could not be verified.

Regarding Berkson’s bias, patients with COVID-19 pneumonia were diagnosed based on radiography findings. The increased access to radiography in COVID-19 cases could increase the chances of detecting PTB in these patients. In contrast, individuals in the population control and negative control groups were devoid of this chance unless they had a lower respiratory symptom or were scheduled for a routine chest radiography before any operation. This bias was likely to reduce after the patients were discharged. Thus, Berkson’s bias could explain only the relatively severe cases in the initial 30-day period of follow-up.

Steroids are well known to impair cell-mediated immunity.52 During hospitalisation, the use of steroids in Thailand was limited only to patients with severe pneumonia. Almost half of them passed away before the follow-up time was sufficient to detect active PTB. Hence, steroid use was unlikely to be a reasonable explanation.

Whether the bacteriologically confirmed PTB cases were newly acquired or reactive cannot be established. While SARS-CoV-2 is different from TB, it probably weakens individuals with TB infection and accelerates their progression toward active PTB. Our study in the pre-COVID-19 period found that PTB was mostly misdiagnosed as unspecified pneumonia.11

During the COVID-19 pandemic, some patients with COVID-19 pneumonia might already be active PTB. However, such cases would be a minority as we advocated the health policy to reduce delayed diagnosis.11 In this study, patients with PTB were diagnosed after being discharged from the hospital in the median of 175 days.

They were more likely to have active PTB during their post-COVID-19 period.
The study’s strength is the completeness of COVID-19 and PTB diagnoses data based on a large population by the law for reimbursement that was followed up for at least 180 days. The major biological risk factors for PTB, i.e., age, sex, HIV infection, DM, cancers, and COPD, were already controlled for. Thus, they could not explain this independent association between COVID-19 and PTB.

For the limitation, we did not control for behaviour and other social variables such as time spent in a crowded area, economic status, and poor respiratory hygiene. While those data were unavailable, our sensitivity analysis revealed a very high E-value of 13.78, which means that the unmeasured confounder must have an explicitly high strength of association with both COVID-19 pneumonia and PTB (risk ratio >13.78) to explain away our findings.

To date, no such confounder has been identified in published literatures. Berkson’s bias has been reduced but not completely eliminated by time-varying Cox’s regression. Although PTB was detected on chest radiographs after SARS-CoV-2 infection, it might have existed long before the pandemic. Temporal sequence of these two infections therefore could not be confirmed. The results must be interpreted with caution.

In conclusion, people who recently survived COVID-19 pneumonia in a TB endemic area should receive special attention regarding the robust association between COVID-19 pneumonia and the hazard of detectable active PTB. Further studies on this population on TB infection and the cost-utility for screening and TB preventive therapy are needed.