Interaction of Glycemic Control and Statin Use on Diabetes-Tuberculosis Treatment Outcome: A Nested Case-Control Study

This study aims to explore the interaction of glycemic control and statin use on the treatment outcomes of pulmonary tuberculosis-diabetes comorbidity (PTB-DM) patients. A nested case-control study was conducted in a tuberculosis patients' cohort. We defined cases as patients who experienced unfavorable outcomes. Glycemic control was estimated at the baseline. Statin use was obtained from medical records. The multivariate logistic regression models were developed, and the interaction table invented by Andersson was adopted to analyze the interaction of glycemic control and statin use on treatment outcomes. A total of 2,047 patients were included in this study. There was a significant interaction between glycemic control and statin use on the treatment outcomes. Patients with good glycemic control and no statin use (OR = 0.464, 95% CI: 0.360–0.623) had a lower risk of unfavorable outcomes than those with poor glycemic control and statin use (OR = 0.604, 95% CI: 0.401–0.734). Patients with good glycemic control and statin use had the lowest risk of unfavorable outcomes (OR = 0.394, 95% CI: 0.264–0.521). Glycemic control in diabetes-tuberculosis treatment should be paid considerable attention. Patients can benefit from statin use even if they have poor glycemic control. Patients with good glycemic control and statin use can have the best outcomes.


Introduction
Tuberculosis (TB) remains to be a major global public health problem.Patients with diabetes carry a 3-fold risk of developing active tuberculosis than those without diabetes [1].Te global estimated prevalence of diabetes among TB patients was 15.3% [2], while in China, it was estimated to be 7.8% [3].Although this level is relatively low, pulmonary tuberculosis-diabetes comorbidity (PTB-DM) is still a serious issue in China because of the large population base.Te PTB-DM patients experienced higher rates of unfavorable outcomes than nondiabetic patients [4,5].
Numerous studies have shown that PTB-DM patients with poor glycemic control have an increased risk of unfavorable outcomes [6,7].Tis is because hyperglycemia impairs innate and adaptive immunity, exacerbating microbial infections [8].Vrieling et al. found that PTB-DM patients have a unique lipid profle in the interaction between DM-induced dyslipidemia and TBinduced changes in lipoprotein metabolism [9], which can explain their increased risk of developing atherosclerosis and cardiovascular diseases.Statins are lipid-lowering agents and are commonly prescribed to patients with diabetes.
Statins reduce cholesterol and triglycerides by blocking 3hydroxy-3-methylglutaryl coenzyme A reductase.Tere is some evidence that cholesterol plays an important role in the maintenance of persistent chronic infection caused by Mycobacterium tuberculosis [10,11].It is indicated that stains can inhibit Mycobacterium tuberculosis by reducing cholesterol [10,11].In addition, statins may exert anti-infammatory, antioxidative stress, immunomodulatory, and anti-infective efects on the host during TB [12].Statins may have positive prognostic efects on patients with tuberculosis.
In conclusion, it has been found that glycemic control and statin use both afect the outcome of antituberculosis treatment.However, the interaction of glycemic control and statin use has rarely been investigated.In this paper, we explored their interaction to provide some recommendations for PTB-DM treatment.

Study Design and Participants.
We conducted a nested case-control study in the Innovation Alliance on Tuberculosis Diagnosis and Treatment (Beijing) (IATB) cohort of TB patients.Beijing Chest Hospital, Tianjin Haihe Hospital, Wuhan Pulmonary Hospital, and Shenyang Tenth People's Hospital joined IATB early.Te cohorts at these hospitals were established early and are composed of patients from all over China.Terefore, we included the cohorts of these four hospitals in our study.
Patients' demographic and clinical information was collected at the baseline and during hospitalization.All the patients in our study underwent DOTS for medication use during hospitalization.After discharge, a professional staf member follows up with patients on their medications and TB-related test results.Te cut-of date for follow-up was the date when the patient's course of antituberculosis treatment was completed (initial treatment at the end of the 6th month and retreatment at the end of the 8th month).Tese data were uploaded into the system developed by the IATB.
In our study, all PTB-DM patients between 1 September 2009 and 1 September 2019 were determined.We defned cases as patients who experienced unfavorable outcomes.Patients who had favorable outcomes were classifed into the control group.Te study fowchart is summarized in Figure 1.

Inclusion Criteria and Exclusion Criteria.
All patients included in our study were diagnosed with PTB-DM.Te diagnosis of PTB was made by a physician according to the national PTB diagnostic criteria.T2DM was defned as participants diagnosed with T2DM by a physician or through a medical record review indicating the use of antihyperglycemic agents.All included patients had sputum specimens retested at a tuberculosis control facility at the end of the last month of the course and at least 30 days before the interval before the end of the last month (for initial tuberculosis patients at the end of the ffth and sixth months and for retreatment patients at the end of the seventh and eighth months).
None of these patients were in pregnancy or lactation.Individuals were excluded if they had been diagnosed with severe respiratory failure, disseminated tuberculosis, malignancy, AIDS, leukemia, or HIV-positive.Any patient who had mechanical ventilation was excluded.Patients whose data were incomplete or the follow-up was less than 6 months were also excluded.

Variables of Interest.
Treatment outcomes were defned based on the World Health Organization guidelines.Favorable outcomes were defned as a PTB patient who was smear-or culture-negative in the last month of treatment and on at least one previous occasion.Unfavorable outcomes were defned as the following: (1) died: a TB patient who dies for any reason before starting or during the course of treatment; (2) treatment failed: a TB patient whose sputum smear or culture is positive at month 5 or later during treatment; (3) lost to follow-up: a TB patient who did not start treatment or whose treatment was interrupted for 2 consecutive months or more; (4) still on treatment: a TB patient who was still on treatment at the time of study termination.
Te level of glycated hemoglobin (HbA1c) was assessed before the start of treatment to estimate glycemic control.HbA1c < 8.0% was defned as good glycemic control; otherwise, poor glycemic control would be recognized [13].
We defned statin use as receiving statin prescriptions continuously for at least 7 days during treatment, as in the previous studies [14,15].Statins in this study included simvastatin, lovastatin, pravastatin, fuvastatin, atorvastatin, rosuvastatin, and pitavastatin.

2.4.
Covariates.Demographic characteristics, including gender, age, and sputum smear results, were collected at the baseline.Te patients included in the study were categorized into three age groups (20-44 years, 45-64 years, and >65 years) [16].Te diagnosis of drug-related liver injury was based on the Expert Recommendations for the Diagnosis and Management of Drug-Related Liver Injury Due to Antituberculosis Drugs [17].Drug resistance was defned based on the results of drug sensitivity tests for Mycobacterium tuberculosis conducted during the patient's hospitalization.
First-line drugs are used as the standard regimen for drug-sensitive TB, and second-line drugs are frequently used in patients with drug-resistant TB.Te frst-line regimen was defned as treatment drugs that include only isoniazid, rifampin, pyrazinamide, streptomycin, or ethambutol.Te second-line regimen was defned as the use of any of the following drugs: injectable antibiotics-aminoglycosides, fuoroquinolones, ethylthioisonicotinamide, prothioisonicotinamide, para-aminosalicylic acid, and cycloserine [18].
Te age-adjusted Charlson comorbidity index (ACCI) was evaluated as defned by Charlson.Te fnal ACCI was determined with 19 medical conditions being considered, with a score range of 1-6 for each comorbidity.For patients over 40 years old, 1 point was added for each decade [19,20].Since all patients were diagnosed with T2DM, the ACCI was not less than 1 point.
Serum C-reactive protein (CRP) and globulin (GLB) levels testing was performed in all included patients.Serum CRP and GLB levels were measured by collecting fasting peripheral venous blood from patients in the early morning.Te results of both protein assays used in this study were obtained before the start of the patients' inpatient treatment.

Quality Control.
Physicians at each hospital have undergone years of training, which ensures the accuracy of the information on the treatment of TB patients.In addition, IATB has trained professionals to manage and verify the data uploaded by each hospital.Tese staf members can provide timely feedback, add necessary information, and correct outliers.

Statistical Analysis.
Te sample size formula for casecontrol studies was used to determine the required sample size [21].We considered a 0.05 two-sided signifcance level, and the statistical power (1-β) was 90%.A sample size of at least 276 patients per group was estimated, requiring a total of at least 552 patients.
Continuous data were described as means with standard deviations (SD) or median (interquartile range, IQR).Differences of continuous data between the groups were analyzed using t-test or Wilcoxon rank-sum test.Qualitative data were described as frequency (percentage), and chi-square test (χ 2 -test) was used to compare the diferences between diferent groups.Te variables with P value of less than 0.10 in univariate analysis were included in multivariable analysis.An unconditional multivariable logistic regression model was used to estimate association between glycemic control and statin use on the treatment outcomes, and the forward selection method was used to select variables for confounder adjustment.Te results were reported as OR with 95% CI.
Te results of this unconditional multivariable logistic regression model and the interaction table invented by Andersson et al. [22] were used to analyze the additive interaction of glycemic control and statin use on the treatment outcomes.Te following interaction indices were calculated: (1) the relative excess risk due to interaction (RERI); (2) the attributable proportion due to interaction (AP); (3) the synergy index (SI).If the confdence interval of RERI and AP contains 0 or the confdence interval of SI contains 1, then there was no interaction between glycemic control and statin use.We defned α � 0.05 as the signifcance level, and two-tailed P ≤ 0.05 was considered to indicate statistical signifcance.SAS version 9.4 (SAS Institute, Gar, NC, USA) and Excel software were used for statistical analysis.

Results
A total of 2,047 PTB-DM patients were consecutively recruited into this study, 1428 (68.72%) had poor glycemic control, and 1601 (77.05%) did not use statins.

Te Outcomes of Antituberculosis Treatment in PTB-DM
Patients.Te incidence of unfavorable outcomes was 27.09%.Tere were signifcant diferences in the incidence of unfavorable outcomes among patients of diferent sex, age, sputum smear results, ACCI levels, drug-related liver injury, treatment regimens, statin use, and glycemic control (P < 0.05, Table 1).Serum CRP and GLB levels were signifcantly higher in patients with unfavorable outcomes than in those with favorable outcomes (P < 0.05, Table 1).

Te Interaction of Glycemic Control and Statin Use on the
Treatment Outcome.Patients were categorized into four subgroups based on their glycemic control and statin use conditions.Tere was an additive interaction between glycemic control and statin use, with RERI, AP, and SI of 0.388 (95% CI: 0.165-0.669),0.914 (95% CI: 0.424-1.437),and 0.786 (95% CI: 0.694-0.954),respectively (Table 3).Patients with good glycemic control and statin use had the lowest risk of unfavorable outcomes.Patients with good glycemic control and no statin use had a lower risk of unfavorable outcomes than those with poor glycemic control and statin use (Figure 2).

Comparison between Diferent Subgroups.
In the four subgroups, the overall trend of changes in the treatment outcomes was statistically signifcant (P for trend <0.05).
Patients with good glycemic control and statin use had the lowest incidence of unfavorable outcomes (14.84%,Table 4).
Te overall trend of changes in the proportion of smearnegative TB was statistically signifcant (P for trend <0.05).Te lowest percentage of smear-negative TB was found in patients with good glycemic control and statin use (35.94%,Table 4).

Discussion
In our study, there was a signifcant interaction between glycemic control and statin use on the treatment outcomes.PTB-DM patients who have poor glycemic control can also beneft from statin use.[5][6][7]23].Our study showed that good glycemic control reduced the risk of unfavorable outcomes regardless of statin use.In our study, glycemic control was assessed based on the HbA1c level at the baseline.HbA1c has been recognized as the primary biomarker for long-term glycemic control in patients with diabetes [24].Previous studies reported that glycemic control assessed based on the HbA1c level at the Chi-square test.2) Wilcoxon rank-sum test was used.Unadjusted.2) Adjusted for sex, age category, sputum smear result, ACCI, drug-related liver injury, anti-TB treatment regimen.3) Adjusted for sex, age category, sputum smear result, ACCI, drug-related liver injury, anti-TB treatment regimen, CRP, and GLB.OR: odds ratio; CI: confdence interval.Canadian Journal of Infectious Diseases and Medical Microbiology baseline was associated with antituberculosis treatment outcomes [6,7].

Comparisons with Other Studies. Te role of glycemic control in antituberculosis treatment has been highlighted in previous studies
In our study, we found that the efect of glycemic control was more prominent in antituberculosis treatment compared to statin use.Magee et al. [7] reported that poor glycemic control was associated with unfavorable outcomes, and glycemic control at the baseline is more important than therapeutic measures.Tis may be attributed to that poor glycemic control impedes the activation of alveolar macrophages, leading to inadequate phagocytosis of Mycobacterium tuberculosis [8,25].Patients with poor glycemic control were also found to have lower levels of IFN-c [26], which is considered to be closely associated with the prognosis of tuberculosis [27].Terefore, glycemic control in diabetes-tuberculosis treatment should be paid considerable attention.
In our study, statins reduced the unfavorable outcome in PTB-DM patients, and this efect persisted even in patients with poor glycemic control.As a result of Mycobacterium tuberculosis infection, the body triggers physiological stress when fghting against the pathogen-induced systemic infammatory state, which negatively afects metabolism and leads to stress hyperglycemia and impaired glucose tolerance, resulting in difcult glycemic control [28].In addition, there are interactions between antituberculosis drugs and glucose control drugs, such as rifampin, which accelerates the metabolism of hypoglycemic drugs other than metformin [29].Te plasma concentrations of these drugs may be signifcantly lower than those of diabetic patients without tuberculosis, resulting in poorer glycemic control in patients.With the infammatory response caused by TB and the infuence of anti-TB drugs, PTB-DM patients have more difculty with glycemic control.Terefore, our study provides possible therapeutic adjuvant drugs for these patients.
Several studies have reported that statins reduce the risk of TB in T2DM patients [15,30], even if the patients have poor glycemic control [31,32].However, epidemiological studies on the association between statin use and antituberculosis treatment outcomes are scarce.It is generally believed that strains have antitubercular efects, which was, nevertheless, denied by Pan et al. [31], who found that statin use was not associated with improved treatment outcomes in TB patients.Te possible reason is that their study was conducted based on the insurance database after adjusting only for demographic characteristics and the Charlson comorbidity index, which may lead to several biases and obscure the potential efects of statins.Te role of statins in the prognosis of infectious diseases has been observed.Douglas et al. [33] found that treatment with statins reduced the mortality of patients with pneumonia.Ghayda's metaanalysis also showed that statin use reduced mortality caused by various infectious diseases [34].Recent studies have  Trend χ 2 test was used. 6 Canadian Journal of Infectious Diseases and Medical Microbiology shown that statins can activate macrophages and enhance their phagocytic function [35,36].In addition, statins have been found to increase IFN-c secretion after TB infection [37].In summary, PTB-DM patients who have poor glycemic control can beneft from statin use.In this study, after adjusting for serum CRP and GLB levels, patients with good glycemic control and statin use had the lowest risk of unfavorable outcomes.Interestingly, we found that the proportion of patients with favorable outcomes was higher in our study in patients with higher ACCI.Further analyses showed that the proportion of patients with poor glycemic control and nonstatin use was lower in patients with higher ACCI than lower ACCI.A survey by the Chinese Centre for Disease Control and Prevention also showed that older people had better glycemic control than younger people [38].Meanwhile, potential drug interactions should not be ignored.A Japanese study showed that sputum transformation time was shorter in patients with lower CRP levels compared to higher CRP levels [39].As an acute phase protein, CRP enhances the phagocytosis of immune cells for pathogens and improves the chemotaxis of various types of immune cells in body.In addition, CRP can refect the severity of TB [39].GLB are immune substances produced by the body after a pathogen infection.After pathogen infection, GLB levels will continue rising if the pathogen cannot be eliminated in time.In a study that included 226 TB patients, higher serum GLB levels were signifcantly associated with treatment outcomes [40].
In our study, the proportion of sputum smear-negative TB in these patients was the lowest.TB patients with smearnegative were found to have higher levels of immune cells and IFN-c [41,42].As mentioned above, both glycemic control and statin use can improve immune functions.With their combination, patients may have further improved immune function and better IFN-c levels.Statins exert a promoting efect on the sterilizing ability of antituberculosis drugs [35,36] and an inhibitory efect on Mycobacterium tuberculosis [43].Terefore, patients with good glycemic control and statin use have better antituberculosis treatment outcomes.

Strengths and Limitations.
Previous studies have focused on the role of statins in tuberculosis development in diabetic patients.In the present study, we explored not only the efect of statin use but also the interaction of glycemic control and statin use on outcomes in PTB-DM patients.Moreover, our study patients were obtained from four large medical institutions, which are located in North, Northeast, and Central China, ensuring the representativeness of the sample.
However, our study has several limitations.First, there were insufcient data available on certain personal characteristics of patients, such as their history of diabetes, lifestyle, medication habits, and level of education, which may also impact the outcome of TB treatment.Second, our analysis only considered the presence or absence of statins and did not take into account the dosage of statins.Furthermore, in our study, the treatment regimens were only classifed by drug into frst-line and second-line treatment regimens.Tird, ATT duration and drug sensitivity patterns were not analyzed in the study.Fourthly, only four large hospitals in China were included in this study, and the results may not be universally applicable because the treatment protocols of hospital and population characteristics may vary in diferent regions of China.Additionally, our study population was limited to China, and the fndings may not be generalizable to other countries or populations due to diferences in treatment protocols and ethnicities.Finally, our study was retrospective, and a certain amount of bias is unavoidable.

Conclusion
Glycemic control in diabetes-tuberculosis treatment should be paid considerable attention.Patients can beneft from statin use even if they have poor glycemic control.Te combination of good glycemic control and statin use contributes to better outcomes in diabetes-tuberculosis treatment.

Figure 1 :
Figure 1: Te study fowchart of the nested case-control cohort based on the IATB cohort.

Table 1 :
Characteristics of PTB-DM patients with diferent treatment outcomes.

Table 2 :
Efect of glycemic control and statin use on treatment outcomes.

Table 3 :
Indexes of additive interaction between glycemic control and statin use on treatment outcomes.

Table 4 :
Comparison of unfavorable outcomes and negative smear between diferent subgroups1).