Effect of complicated, untreated and uncontrolled diabetes and pre‐diabetes on treatment outcome among patients with pulmonary tuberculosis

Patients with tuberculosis and diabetes have a higher risk of unfavourable anti‐tuberculosis treatment outcomes. In the present study, we aimed to evaluate the effects of various diabetes statuses on the outcomes of patients with pulmonary tuberculosis.


INTRODUCTION
Tuberculosis (TB) remains a global health burden since the coronavirus disease 2019 pandemic.Integrated patientcentred care and prevention are key components of the World Health Organization's End TB Strategy for TB patients with other health conditions.Diabetes mellitus (DM) is an important risk factor for active TB and poor treatment outcomes.In 2020, an estimated 370,000 new cases of TB were attributable to DM. [1][2][3] Globally, over 15% of people with TB have DM, compared to 9.3% of the general population. 1 This means approximately 1.5 million people with both TB and DM require coordinated care and follow-up for optimal management.The global DM cases are expected to increase by 50% from 2019 to 2045. 4 This dual epidemic requires early and decisive action to ensure that TB is a public health priority. 5n the Republic of Korea, the prevalence of DM was 16.7% among adults in 2020, and only 65.8% of adults with DM were aware of their disease and treated with antidiabetic medications. 6Furthermore, only 24.5% of adults with DM achieved optimal glycemic control.The incidence of TB in Korea was 44.6 per 100,000 people in 2021, 7 and the increase in older adults with TB suffering from comorbidities, such as diabetes, is a major obstacle to TB control. 80][11] However, the effect of glycemic control on TB treatment remains unclear.Limited data and inconsistent findings exist regarding the effects of glucose control on TB treatment outcomes. 12,13In the present study, we evaluated the effects of various diabetes statuses on treatment outcomes among patients with pulmonary TB using a national TB registry database and a multicentre prospective cohort database.

A national TB registry database
In Korea, a public-private mix TB control project was expanded nationwide in 2011.All patients are monitored and followed up until treatment completion by TB specialist

SUMMARY AT A GLANCE
The effect of various stages of diabetis mellitus (DM) on the outcomes of patients with tuberculosis (TB) was analysed.Untreated and complicated DM at the time of TB diagnosis increased the risk of unfavourable outcomes.
nurses.We constructed the Korean Tuberculosis Cohort (KTBC) database, a prospective observational registry database of notified patients with TB within the project.Clinical and epidemiological data were collected at the time of TB diagnosis and treatment completion.For this study, we enrolled patients with rifampicinsusceptible pulmonary TB who were notified between July 2018 and December 2020 and identified those with a DM history.Patients with DM were further categorized into those with and without complications, such as cerebrovascular disease, coronary artery disease and nephropathy.

A multicentre prospective cohort of pulmonary TB
The Cohort Study of Pulmonary Tuberculosis (COSMOTB) was a multicentre prospective observational study conducted in Korea between 2019 and 2021.The main objective of this cohort study was to describe the clinical outcomes of patients with pulmonary TB and to identify the risk factors for unfavourable outcomes.We enrolled adults aged ≥19 years with pulmonary TB from 18 universityaffiliated hospitals.All patients were regularly followed up for 24 months after the initiation of anti-TB treatment.Patients with rifampicin-resistant TB were excluded from the analysis.We collected data on random blood glucose levels, glycated haemoglobin (HbA1c) levels and DM history.Patients were classified as having DM if one of the following criteria was met: (1) random glucose level ≥200 mg/dL, (2) HbA1c ≥6.5% or (3) previous DM history.Patients with diabetes were further categorized into (1) those treated with and without glucose-lowering medications at the baseline, (2) those with and without diabetes complication at the baseline (random glucose level ≥180 mg/dL, HbA1c ≥6.5% or presence of end organ damage) and (3) those with and without uncontrolled follow-up random glucose level (≥180 mg/dL).The end organ damage of diabetes was defined as nephropathy, retinopathy, neuropathy, cerebrovascular disease and coronary artery disease in the COSMOTB.Among patients without previous DM history, we also identified those with pre-diabetes if one of these criteria were met: (1) random glucose level ≥140 but <200 mg/dL or (2) HbA1c ≥5.7 but <6.5%.

Definition of anti-TB treatment outcome
The primary outcome was an unfavourable treatment outcome, a composite of death, failure, loss to follow-up, still-on-treatment and transfer-out, which were defined according to the Korean TB guidelines adopted from the World Health Organization. 14The secondary outcome was all-cause mortality.Patients who continued anti-TB treatment after 1 year were defined as still-on-treatment.

Data collection
Factors that may influence treatment outcomes were: sex, age, BMI, malignancy, prior TB history, extrapulmonary involvement, initial TB-related symptom, cavitation on chest radiography and sputum acid-fast bacilli smear test results.A predefined checklist was implemented to collect initial TB-related symptoms.We combined the chest radiographic findings and sputum smear test results to generate a single variable with two strata: mild disease (non-cavitary disease and negative smear result) and severe disease (cavitary disease or positive smear result).

Statistical analysis
Continuous variables are presented as medians and interquartile ranges, whereas discrete variables are presented as frequencies and percentages.The baseline characteristics of patients with and without DM were compared, and univariable analyses were performed using the chi-square test for categorical variables and the Mann-Whitney U test for continuous variables.For regression analysis, we performed a univariable analysis to evaluate the association between DM and outcomes.Subsequently, we selected variables with p-value <0.20 and performed a multivariable logistic regression analysis.Statistical significance was set at p-value <0.05.All statistical analyses were performed using the SPSS software (version 17.0; Statistical Product and Services Solutions, Chicago, IL, USA).
Subgroup analyses were performed to assess the effects of complicated, untreated and uncontrolled DM on outcomes with reference to those without DM.Patients with complicated DM were identified from the KTBC database.Using the COSMOTB database, we selected three stratification strategies and identified patients with untreated DM, complicated DM at baseline and uncontrolled DM at the follow-up visits.We also evaluated the effects of prediabetes on outcomes in patients without DM using the COSMOTB database.For the sensitivity analysis, we identified cases of treatment failure and death to redefine the unfavourable outcomes.Further multivariable logistic regression analyses were conducted to assess the effect of various glycemic status on treatment failure and death.

Characteristics of enrolled patients with pulmonary TB
The prevalence of TB-DM in the KTBC and COSMOTB were 21.5% and 30.3%, respectively.(Table 1 and Figure S1 in the Supporting Information).Patients with pulmonary TB-DM were older, predominantly male, had a higher BMI, and a higher malignancy rate.Proportion of patients with initial severe disease was significantly higher in the TB-DM group.The baseline characteristics of patients with TB-DMin the COSMOTB group were similar to those of patients in the KTBC group.

Effect of glycemic status on treatment outcome
Among the 2966 patients with DM in the KTBC, 693 (23.4%) patients had DM complications.Among the patients with complicated DM, 368 (53.1%) and 172 (24.8%) had unfavourable outcomes and all-cause mortality, respectively (Figures 1 and 2, and Table S1 in the Supporting The associations between complicated DM and unfavourable outcome (aOR, 1.96, 95% CI, 1.67-2.30)and all-cause mortality (aOR, 2.11; 95% CI, 1.75-2.57)were slightly strengthened in terms of the effect estimate.
In the sensitivity analysis, DM was significantly associated with treatment failure and death in both KTBC (aOR, 1.37; 95% CI, 1.20-1.57)and COSMOTB (aOR, 2.71, 95% CI; 1.52-4.90)databases (Table S2 in the Supporting Information).The analysis revealed a significant association of treatment failure and death with complicated and untreated DM, but not with uncontrolled DM and prediabetes (Table S3 in the Supporting Information).

DISCUSSION
This is the first large-scale prospective study to comprehensively investigate the impact of various stages of DM on anti-TB treatment outcomes.By analysing data from the national registry database, we revealed that complicated DM can adversely affect treatment prognosis.Furthermore, we analysed prospective multicentre cohort data using the same analytical methods, thereby confirming our results that both complicated and untreated DM have detrimental effects on outcomes.Despite increasing awareness of the need for multidisciplinary collaboration to combat TB-DM, there is scant evidence to provide clinical guidance for optimal management of uncontrolled and complicated DM at the time of TB diagnosis.In a Korean retrospective study, uncontrolled DM at baseline was an independent risk factor for treatment failure or death. 9In a Taiwanese retrospective study, patients with DM-related comorbidities had an increased risk of unfavourable outcome. 15In our study, complicated DM was defined as poor glycemic control or complications at initial assessment.The higher incidence of unfavourable outcomes could be ascribed to a higher mortality rate due to DM-related comorbidities, such as cardiovascular disease. 11n addition, a higher risk of drug-drug interactions, adverse events and high pill burdens could impose hardship on TB-DM management, 16 which, in turn, slows treatment response and increases the number of clinically or microbiologically failed cases.
Another key finding of our study was that TB patients with untreated DM at the initial assessment had worse treatment outcomes than those without DM.The COSMOTB showed higher prevalence of DM compared to the KTBC but aligned with recent Korean studies, 17 emphasizing the need for accurate health assessments to check concomitant DM.Using prospective multicentre cohort data, we stratified various DM statuses and assessed their impact on treatment outcomes.In this cohort, patients who did not receive any anti-diabetic management at the time of TB diagnosis constituted 4.1% of the total cohort and 13.4% of patients with DM.According to recent guidelines, this result has two clinical implications. 18First, routine screening of DM should be performed in patients with TB at the time of diagnosis, however; few countries have incorporated it into their national TB control programs.Second, people with blood glucose levels significantly higher than the diagnostic threshold for DM require appropriate evaluation by a diabetologist.Patients with uncontrolled DM during follow-up had higher proportions of unfavourable outcomes and mortality in our study; however, statistical significance was not achieved.A recent systematic review suggested that glycemic control may have a favourable effect on anti-TB treatment outcomes. 12However, the authors could not draw a firm conclusion because of the high methodological and clinical heterogeneity of the studies and highlighted the importance of conducting future research using a standard design and analysis plan.For example, we used random blood glucose levels as a surrogate marker of poor glycemic control in COSMOTB.Repeated testing of HbAlc at the end of the intensive phase and during the continuation phase of anti-TB treatment could more accurately assess the glycemic status.
Although we did not find an association between prediabetes and anti-TB treatment outcomes, prediabetes is increasingly recognized as an important metabolic state that renders individuals at a high risk of developing overt diabetes.Prediabetes might be composed of transient stressinduced dysglycemia, and its adverse effects on the immune response and treatment outcomes might be less than those of chronic DM. 10 However, people with prediabetes at the time of TB diagnosis are still at a higher risk of developing overt DM and should be counselled to identify modifiable risk factors and follow healthy lifestyle advice.
Decisions regarding TB treatment and management of coexisting health conditions should be personalized, considering disease severity, affected organs and treatment response.For patients with uncontrolled diabetes, TB treatment can be extended to 9 months due to higher recurrence rates. 19The current study did not analyse post-treatment recurrence rates or the impact of treatment duration on outcomes; thus, it lacks evidence for extending TB treatment for DM patients.However, we revealed that complicated and untreated DM led to more adverse outcomes.Combining stringent glucose control with prolonged anti-TB treatment might benefit these patients, but further research is essential to confirm its efficacy.
Our study has several limitations.First, because the KTBC and COSMOTB are both observational cohorts, T A B L E 3 Association between diabetes and all-cause mortality among enrolled patients with pulmonary tuberculosis: (A) a national tuberculosis registry database (Korea Tuberculosis Cohort, KTBC) and (B) a multicenter prospective cohort of pulmonary tuberculosis (COSMOTB).unmeasured or residual confounding factors might have existed.Second, clinically differentiating true causes of cerebrovascular disease, coronary artery disease and nephropathy among patients with diabetes can be difficult.When defining the diabetic complications, we first tried to exclude other possible causes.We could not collect data on some complications, such as retinopathy and nephropathy, in the KTBC database.However, we used the chart review version of the Charlson Comorbidity Index to capture the presence of retinopathy and neuropathy in the COSMOTB database.We verified consistent findings between the two databases, supporting their reproducibility.Third, we could not implement a detailed DM-related questionnaire in the COSMOTB.

Variables
Based on numerous studies revealing that complications are significantly high in patients with uncontrolled blood glucose level, we employed HbA1c and random blood glucose results to complement the definition of complicated DM.Fourth, anti-TB treatment outcomes between the two databases varied due to different characteristics of the participants.While KTBC encompasses all notified Korean patients, COSMOTB includes patients from university-affiliated hospitals who consented to the study.The COSMOTB showed lower death rates, possibly because of excluding critically ill or mentally impaired individuals, while transfer-out rates were higher in KTBC, potentially linked to the patients' choices after universal health coverage in Korea.These disparities underscore the need for careful consideration when interpreting our findings.
In conclusion, our study confirms the deleterious effects of untreated and complicated DM on anti-TB treatment.It is important to screen for DM status and complications at baseline and incorporate proactive DM management.

F I G U R E 1
Effect of complicated, untreated and uncontrolled diabetes and pre-diabetes on unfavourable treatment outcomes in patients with pulmonary tuberculosis: (A) Percentages of unfavourable outcome; (B) a national tuberculosis registry database (Korea Tuberculosis Cohort, KTBC); and (C) a prospective cohort study of pulmonary tuberculosis (COSMOTB).KTBC, Korean Tuberculosis Cohort; COSMOTB, Cohort Study of Pulmonary Tuberculosis; DM, diabetes mellitus.All odds ratios (ORs) were calculated with reference to non-diabetes, except for pre-diabetes, which was compared with non-pre-diabetes.Variables with p-value <0.20 in the univariable analyses were selected for the multivariable logistic regression analysis.FI G U R E 2 Effect of complicated, untreated and uncontrolled diabetes and pre-diabetes on all-cause mortality in patients with pulmonary tuberculosis: (A) percentages of mortality; (B) a national tuberculosis registry database (Korea Tuberculosis Cohort, KTBC); and (C) a prospective cohort study of pulmonary tuberculosis (COSMOTB).KTBC, Korean Tuberculosis Cohort; COSMOTB, Cohort Study of Pulmonary Tuberculosis; DM, diabetes mellitus.All odds ratios (ORs) were calculated with reference to non-diabetes, except for pre-diabetes, which was compared with non-pre-diabetes.Variables with p-value <0.20 in the univariable analyses were selected for the multivariable logistic regression analysis.
T A B L E 1 Baseline characteristics of enrolled patients with pulmonary tuberculosis in a national tuberculosis registry database (Korea Tuberculosis Cohort, KTBC) and a multicenter prospective cohort of pulmonary tuberculosis (COSMOTB).
T A B L E 2 Association between diabetes and unfavourable treatment outcomes among enrolled patients with pulmonary tuberculosis: (A) a national tuberculosis registry database (Korea Tuberculosis Cohort, KTBC) and (B) a multicenter prospective cohort of pulmonary tuberculosis (COSMOTB).
Note: Variables with p-value <0.20 in the univariable analyses were selected for the multivariable logistic regression analysis.Abbreviations: COSMOTB, Cohort Study of Pulmonary Tuberculosis; KTBC, Korean Tuberculosis Cohort; TB, tuberculosis.