Reduced mortality from lower respiratory tract disease in adult diabetic patients treated with metformin

Abstract

BACKGROUND AND OBJECTIVE:

Chronic lower respiratory diseases (CLRD) increase the risk of type 2 diabetes, which in turn may worsen lung function. Metformin, a common antidiabetic with anti-inflammatory and antioxidant properties, may improve respiratory outcomes. Therefore, we examined the association of metformin use with the risk of mortality from CLRD.

METHODS:

We analyzed data on 1988–1994 and 1999–2010 National Health and Nutrition Examination Surveys participants aged 40 years or older who had diabetes and were followed for mortality through 2011. Information on prescription medicine was collected at baseline and CLRD-related mortality during follow-up was defined using the 10th Revision of the International Classification of Diseases (ICD-10). Cox proportional hazards modeling was used to determine the mortality hazard ratio (HR) associated with metformin use, adjusting for relevant covariates.

RESULTS:

A total of 5,266 participants with a median follow-up of 6.1 years were included. The prevalence of metformin use was 31.9% and 1,869 participants died during follow-up, including 72 of CLRD. In the adjusted Cox proportional regression analysis, metformin was associated with a decreased risk of CLRD mortality in the overall population (HR: 0.39, 95% CI: 0.15–0.99) and among participants with baseline CLRD (HR: 0.30, 95% CI: 0.10–0.93), after adjusting for age, gender, race/ethnicity, cigarette smoking, body mass index, current asthma and COPD, insulin and other diabetic medications, and glycohemoglobin level. We found no association between other antidiabetic medications and CLRD mortality.

CONCLUSION:

In this sample representative of the U.S. population, metformin was associated with lower CLRD mortality in adults with diabetes.

Summary at Glance

This population-based prospective cohort demonstrated that metformin use in type 2 diabetes was associated with a significant decrease in the risk of mortality from chronic lower respiratory disease after adjusting for relevant covariates. This reduction was even more pronounced among subjects with chronic lower respiratory disease at baseline.

INTRODUCTION

Chronic lower respiratory diseases (CLRD) like asthma and chronic obstructive pulmonary disease (COPD) are a major cause of morbidity and mortality, and adults with type-2 diabetes (T2D) are at higher risk for these conditions.^1,2 Hyperglycemia-induced oxidative stress and inflammation, glycosylation of pulmonary proteins, thickening of basal laminae, and susceptibility to respiratory infections could all contribute to pulmonary function impairment in diabetes.³ Similarly, subjects with COPD and asthma are at higher risk of developing T2D.⁴ Metformin, a drug that reduces insulin resistance and is commonly used in the treatment of T2D, acts through complex mechanisms involving AMP-activated protein kinase (AMPK) activation and the inhibition of mitochondrial respiration.³ In addition to its antidiabetic effects, metformin may have anti-inflammatory and antioxidant properties and may thus improve pulmonary outcomes in diabetes.^3,5

Previous studies have examined the association of metformin use with the development of asthma and/or COPD as well as the morbidity related to these conditions in diabetic people.^5–9 However, it is not known whether metformin use reduces CLRD mortality among in people with T2D. Therefore, we tested this hypothesis in a representative sample of the U.S. population with diabetes.

METHODS

Data Source

We used data from the National Health and Nutrition Examination Survey (NHANES) conducted between 1988 and 1994 (NHANES III) and from 1999 to 2010. These correspond to all the NHANES cycles for which mortality data were available. No NHANES survey was available from 1994 (the end on the NHANES III cycle) to 1999 (beginning of NHANES 1999–2000 cycle). The NHANES is a continuous cross-sectional survey of the U.S. noninstitutionalized civilian population done by the National Center for Health Statistics (NCHS) of the Centers for Disease Control and Prevention (CDC) to assess the health of the non-institutionalized civilian U.S. population. It uses a complex multistage sampling design to derive a sample representative of the American population and collects data on participants’ health through interviews, physical examinations as well as laboratory tests. Details on NHANES methods and procedures are available at http://www.cdc.gov/nchs/nhanes/survey_methods.htm.

Participants were classified as having diabetes if they reported a diagnosis of diabetes by a health professional, were taking antidiabetic drugs, had a glycohemoglobin (hemoglobin A1C) ≥6.5%, or fasting plasma glucose ≥126 mg/dL. To ensure our study only included participants with T2D, those who were only taking insulin and no oral antidiabetic drugs were excluded, since they were more likely to have type 1 diabetes (T1D). During the study period, 5,266 adults aged 40 years or older who were identified as having T2D also had data on mortality during follow-up (Figure 1).

Figure 1:

Flow diagram of the participants included in the study

NHANES protocols were approved by the Institutional Review Boards of the NCHS and the CDC and all NHANES participants provided informed consent. Details of the IRB approval are available at http://www.cdc.gov/nchs/nhanes/irba98.htm.

Metformin Use

At baseline, NHANES participants were asked about prescription medicines in the past 30 days during home interviews. For participants who reported taking any medication, the names of up to 20 prescription drugs were recorded from the medication container label. When the container was not available, the drug name was recorded as reported by the participant. The name of each medication was linked to its generic drug name in the Multum Lexicon Plus database to be converted into the standard medication name. Documented antidiabetic drugs included metformin, insulin, glitazones, sulfonylureas, and other oral drugs (i.e. acarbose, glucagon, miglitol, meglitinide).

CLRD Mortality

All NHANES adult participants were followed for mortality through December 31, 2011; we merged the baseline NHANES files with the corresponding follow-up Mortality-Linked Files created by the NCHS. NHANES participants were matched to the National Death Index (NDI) records and death certificates were used for confirmation that the NDI deaths corresponded to the correct subjects. The specific causes of mortality were defined using a standardized list of 113 causes according to the Tenth Revision of the International Classification of Diseases (ICD-10). CLRD mortality was defined as death from asthma (Code J45-J46), emphysema (J43), bronchitis (chronic or other) (J40-J42), and other chronic lower respiratory diseases (J44 and J47).

Covariates

Data on age, gender, race/ethnicity, family income, cigarette smoking, and medical conditions were collected using questionnaires. The family poverty income ratio (PIR) was estimated using guidelines and adjustment for family size, year, and state. Body mass index (BMI) was calculated as weight in kilograms divided by height in meters squared (kg/m²). Participants were considered to have a normal weight if they had a BMI < 25 kg/m², to be overweight if they had a BMI between 25kg/m² and 30 kg/m², and to be obese if their BMI was ≥ 30 kg/m². CLRD at baseline was defined as current asthma (participant with a previous asthma diagnosis who still has asthma) and/or report of a diagnosis of chronic bronchitis or emphysema (COPD) as reported during the NHANES. Other covariates included insulin use, the use of other oral antidiabetic drugs, and A1C levels at baseline. ‘Missing covariates were coded as ‘missing’ to account for the possibility that those study participants with missing data may be different from those with complete data.

Statistical Analysis

Descriptive analyses were performed to summarize the characteristics of study participants by metformin use status. Differences in characteristics between the groups were assessed using chi-square test for categorical variables and independent t-tests for continuous variables. Cox proportional hazards regression modeling was used to estimate the hazard ratio (HR) and corresponding 95% confidence interval (CI) for the risk of CLRD mortality associated with metformin use. For further details on the statistical models applied, see Supplementary Appendix S1. Analyses were performed in SAS (Version 9.4) and the figures were created in STATA (Version 14). P-values < .05 were considered statistically significant in all analyses.

RESULTS

Our sample consisted of 5,266 NHANES participants with a median age of 61 (interquartile range [IQR]: 51 – 70) years with T2D, who were followed for a median of 6.1 (IQR: 3.2 – 10.1) years (see Table 1). During follow-up there were 1,869 recorded deaths (35.5% of participants), including 72 participants who died of CLRD (CLRD mortality was thus 1.42% of the total population, and 3.85% of total deaths). Approximately 50.4% of the sample consisted of females, and 15.8% had a PIR ≤1; 54.3% were smokers, 29.3% were overweight, and 56.7% were obese. Median glycohemoglobin was 6.77.

Table 1:

Baseline characteristics of the study participants, NHANES III, NHANES 1999–2010 (N=5,266)

Characteristics	All participants	No metformin	Metformin	P-value
All participants, %	100	68.1	31.9
Age (years), median (SE)	61.13 (0.40)	61.83 (0.59)	59.91 (0.55)	0.009
Gender, %				0.45
Men	49.6	50.1	48.6
Women	50.4	49.9	51.4
Race/ethnicity, %				0.46
Non-Hispanic Whites	65.1	65.8	63.6
Non-Hispanic Blacks	15.8	15.7	16.0
Hispanic	12.0	11.4	13.3
Other	7.1	7.1	7.1
PIR, %a				0.99
≤1	15.8	15.8	15.9
>1	84.2	84.2	84.1
BMI, %b				<0.0001
<25 kg/m2	14.0	15.0	11.8
≥25 & <30 kg/m2	29.3	31.2	25.3
≥30 kg/m2	56.7	53.8	62.9
Cigarettes smoking, %c	54.3	56.2	49.6	0.0036
Current asthma, %d	8.6	7.5	10.9	0.0038
COPD, %e	13.2	13.4	12.7	0.63
Statins, %	2.9	2.3	4.4	0.0001
Sulfonylureas, %	23.8	22.5	26.8	0.027
Glitazones, %	13.3	9.2	22.0	<0.0001
Other oral antidiabetics, %	1.1	1.1	1.2	0.93
Insulin, %	10.6	8.1	15.9	<0.0001
A1C, median (SE)f	6.77 (0.02)	6.68 (0.03)	6.76 (0.06)	0.38
CLRD Mortality, %	1.4	1.8	0.6	<0.0001

Abbreviations: BMI, body mass index; CLRD, chronic lower respiratory disease; A1C, glycohemoglobin; PIR, poverty-income-ratio; SE, standard error.

P-values from chi-square tests for the difference between metformin and no metformin groups.

Number of participants with missing data:

^aPIR = 578,

^bBMI = 425,

^ccigarette smoking = 5,

^dasthma = 75,

^eCOPD = 12,

^fA1C = 477.

Among the study participants, 1,680 (31.9%) were taking metformin (Table 1). Compared to participants not taking metformin, those who reported taking the drug were slightly younger, although the difference was small (median 59.9 years vs. 61.8 years); had a lower prevalence of cigarette smoking; and had a higher prevalence of using of statin drugs, sulfonylureas, glitazones, insulin, and other oral antidiabetics. However, they had a higher prevalence of obesity and current asthma than those not using the drug. There was no significant difference in sex, race, PIR, COPD, or A1C levels between the two groups. In this unadjusted analysis, the prevalence of CLRD mortality during follow-up was markedly lower among the participants taking metformin than in those not taking the medication (0.6% vs. 1.8%, P<0.0001).

Metformin and CLRD Mortality

In the Cox proportional hazards analysis, metformin use was associated with a significant reduction in the risk of CLRD mortality in the overall sample (HR: 0.39, 95% CI: 0.15–0.99, P=0.049), after adjusting for age, gender, race/ethnicity, cigarette smoking, body mass index, current asthma and COPD, insulin and other diabetic medications, and glycohemoglobin level. To check for potential healthy user bias, we performed a sensitivity analysis limited to participants who reported a diagnosis of diabetes by health professional and the HR improved to 0.36, 95% CI: 0.13–0.98, P=0.046. In another sensitivity analysis, we excluded subjects with GFR < 30 ml/min/1.73m², since metformin is contraindicated below that level, and found very similar results (HR: 0.38, 95% CI: 0.15–1.00, P=0.051). Moreover, metformin was also associated with lower CLRD mortality when the analysis was restricted to participants diagnosed with CLRD at baseline (HR: 0.30, 95% CI: 0.10–0.93, P=0.037) (Figure 2).

Figure 2:

Kaplan Meier curve for the cumulative mortality from CLRD by metformin use in the overall sample (A) and also when the analysis was restricted to participants diagnosed with CLRD at baseline (B). Hazard ratio (HR) for the mortality associated with metformin use was calculated using Cox proportional regression analysis. Models adjusted for age, gender, race/ethnicity, cigarette smoking, body mass index, current asthma, COPD, statins, glitazones, sulfonylureas, other oral antidiabetic drugs (i.e. acarbose, glucagon, miglitol, and meglitinide), insulin, and glycohemoglobin.

In our interaction analysis, there was no significant effect of BMI on the association between metformin use and CLRD mortality. We also found no association between CLRD mortality and the use of sulfonylureas, glitazones, other antidiabetic medications, or insulin either in the whole sample or in participants with CLRD at baseline (data not shown).

DISCUSSION

We report for the first time that metformin use among diabetic adults is associated with decreased risk of mortality from CLRD in a large sample representative of the U.S. population. Among all subjects, there was a significant reduction in the risk of CLRD mortality and in those who had CLRD at baseline, the risk reduction was even more pronounced.

Consistent with our findings, metformin has been suggested to improve other clinical outcomes in COPD. In a small study of 17 patients with a BMI ≥ 25 Kg/m² who had T2D or glucose intolerance and moderate to severe COPD, six months of metformin treatment improved dyspnea, exercise tolerance, and quality of life, but not spirometry or exhaled nitric oxide.¹¹ In a retrospective study in 130 older patients with hospitalized with diagnostic codes for T2D and COPD, Hitchings et al. found that although that the drug was associated with a minor elevation of lactate concentration, the participants taking it had a significantly better overall survival.¹² In the U.S., healthcare utilization over a two-year period in relation to metformin use was studied among 11,260 Medicare recipients between 2007 and 2010 who had diabetes and COPD: metformin was associated with fewer COPD-related emergency room visits and hospitalizations in participants with low COPD severity, and with less overall emergency room visits and hospitalizations regardless of COPD severity.¹³ More recently, Tseng reported that the use of metformin was associated with a ~40% lower risk of COPD.⁶ Yet, not all results are consistent: a recent multicentre, randomized, double-blind, placebo-controlled trial found no effect of metformin administration on glucose elevations during COPD exacerbations, inflammation, or COPD assessment test.⁸ The trial was however small, including 52 participants randomized to receive either metformin (34 participants) or placebo (18 participants). Moreover, these were subjects who did not have diabetes, but were hospitalized for COPD exacerbations. Metformin was administered for one month and the authors recognized that the study was underpowered to detect significant changes in the COPD assessment test.⁸ Similarly, there is emerging evidence that metformin use might be associated with favorable outcomes in asthma. It has been suggested that metformin use could also reduce the risk of developing asthma in diabetic patients, and in a Taiwanese study of 1,332 participants with asthma and diabetes, participants taking metformin had fewer asthma exacerbations and asthma-related hospitalizations.⁷

There are several mechanisms by which metformin might reduce CLRD mortality. Metformin can exert anti-inflammatory properties via the activation of AMPK, which downregulates nuclear factor (NF)-κB. This inhibits the production of inflammatory cytokines and mediators such as interleukin 6, tumor necrosis factor α, fibrinogen, and adhesion molecules (E-selectin, Vascular Cell Adhesion Molecule −1, and Intercellular Adhesion Molecule −1).^3,14,15 Although we found no effect modification by BMI, it has been reported that in obese mice, the drug alleviates eosinophilic inflammation and regulates the levels in bronchoalveolar lavage fluid of chemokines such as eotaxin involved in eosinophils recruitment.¹⁴ Metformin might also reduce airway smooth muscle proliferation induced by platelet-derived growth factor, but this effect has been inconsistently reported in the literature.^16,17 Coherent with the antioxidant properties reported for metformin, the drug could diminish the activity of NADPH oxidase, a major source of reactive oxygen species (ROS).¹⁸ It can also modulate factors involved in the production of ROS from mitochondrial source, while increasing the expression of a selenoprotein that protects leukocytes against oxidative stress through hydrogen peroxide reduction.¹⁹ Another mechanism by which metformin could prevent CLRD death is by decreasing insulin resistance and reducing hyperglycemia which, in T2D, upregulates proinflammatory cytokines and promotes oxidative stress to worsen asthma and COPD outcomes.^3,10,20,21 Furthermore, metformin may impede airway glucose permeability and reduce hyperglycemia-induced pulmonary infections.¹⁰

Our study had limitations. Prescription drug use was available at baseline and not during follow-up. Observational studies like this one are vulnerable to a healthy user bias (i.e., participants taking metformin might be more likely to engage in healthier behaviors that may influence mortality than those not taking the drug). We also cannot completely rule out the possibility of confounding by indication, and thus one might argue that the beneficial effect of metformin on the risk of CLRD mortality could be explained by a lower disease burden in participants taking metformin compared to those not taking the drug or to those taking insulin or other oral antidiabetic medications. However, we did not find such evidence at baseline: there were no differences in COPD prevalence or hemoglobin A1C levels between the groups at enrollment, and in fact we found a higher prevalence of asthma and obesity in the metformin users than the non-metformin users. Moreover, to reduce confounding by indication, we adjusted our analysis for baseline hemoglobin A1C and for the concurrent use of insulin, glitazones, sulfonylureas, and other antidiabetic medications. Furthermore, our sensitivity analysis limited to participants who reported a diagnosis of diabetes by a health professional resulted in stronger results. Our study did not capture the actual use and adherence to the studied medications. We could not examine other interventions that could have prevented CLRD outcomes such as smoking cessation, pulmonary rehabilitation, vaccination status, or the use of macrolides. To prevent participant re-identification, the NCHS perturbed cause and time of death for a selected number of individuals within the publicly-available dataset. Although this does not affect overall population estimates, it is unclear whether it could have affected our results. Despite efforts to limit our study to participants with T2D, it is possible that we may have included participants with T1D. T1D and T2D have different pathogeneses and might affect CLRD outcomes distinctly or through different mechanisms. Some studies have reported a negative association between T1D and asthma,^22,23 while others have reported increased risk of asthma and asthma exacerbations among subjects with T1D.^24–26

In conclusion, we showed that metformin was associated with lower CLRD mortality in a population-based sample of adults with diabetes. Future, prospective studies should focus on accounting for potential sources of bias and confounding, including measures of diabetes control and severity not included in NHANES. If these results are confirmed, they could have public health implications, given the prevalence of diabetes, the burden of CLRD mortality, and the relatively low cost of metformin.