A functional macrophage migration inhibitory factor promoter polymorphism is associated with reduced diffusing capacity

Abstract

Cigarette smoke exposure is the leading modifiable risk factor for chronic obstructive pulmonary disease (COPD); however, the clinical and pathologic consequences of chronic cigarette smoke exposure are variable among smokers. Macrophage migration inhibitory factor (MIF) is a pleiotropic cytokine implicated in the pathogenesis of COPD. Within the promoter of the MIF gene is a functional polymorphism that regulates MIF expression (-794 CATT_5–8 microsatellite repeat) (rs5844572). The role of this polymorphim in mediating disease susceptibility to COPD-related traits remains unknown. We performed a cross-sectional analysis of DNA samples from 641 subjects to analyze MIF-794 CATT_5–8 (rs5844572) polymorphism by standard methods. We generated multivariable logistic regression models to determine the risk of low expressing MIF alleles for airflow obstruction [defined by forced expiratory volume in 1 s (FEV₁)/forced vital capacity ratio <0.70] and an abnormal diffusion capacity [defined by a diffusion capacity for carbon monoxide (DL_CO) percent predicted <80%]. We then used generalized linear models to determine the association of MIF genotypes with FEV₁ percent predicted and DL_CO percent predicted. The MIF-794 CATT₅ allele was associated with an abnormal diffusion capacity in two cohorts [odds ratio (OR): 9.31, 95% confidence interval (CI): 1.97–4.06; and OR: 2.21, 95% CI: 1.03–4.75]. Similarly, the MIF-794 CATT₅ allele was associated with a reduced DL_CO percentage predicted in these two cohorts: 63.5 vs. 70.0 (P = 0.0023) and 60.1 vs. 65.4 (P = 0.059). This study suggests an association between a common genetic polymorphism of an endogenous innate immune gene, MIF, with reduced DL_CO, an important measurement of COPD severity.

INTRODUCTION

Chronic obstructive pulmonary disease (COPD) is a heterogeneous disorder (8). While chronic cigarette smoke (CS) exposure is the leading modifiable risk factor for COPD in the United States, not all smokers develop disease. Among susceptible individuals, the pulmonary manifestations of COPD are manifold and include chronic bronchitis, parenchymal destruction (emphysema), small airway disease, and vascular dysfunction. Similar to overall disease severity, individual COPD-related traits can manifest varying degrees of severity, e.g., two patients may have similar degrees of airflow obstruction, but one may have significantly more emphysema (2). This COPD heterogeneity is influenced by a complex relationship between genetic factors and environmental exposures. Despite advances in COPD genetics, much of the heritability of COPD remains unexplained and there is no cure for this disease (1, 23, 40). However, ongoing high-throughput genomic and candidate gene studies may help identify potential therapeutic targets that underlie susceptibility to COPD-related traits.

Macrophage migration inhibitory factor (MIF) is an immunoregulatory and tissue protective cytokine implicated in the pathogenesis of COPD (14, 33, 34). MIF is secreted in response to microbial, oxidative, and genotoxic stress, and secreted MIF modulates the activity of pleiotropic cellular functions that include proinflammatory and prosurvival signaling cascades. MIF is differentially expressed in subjects with COPD, and murine studies suggest that MIF has divergent roles in the pathogenesis of COPD (14, 22, 34). In studies of asthma and airway disease, MIF promotes airway hyperresponsiveness, bronchial inflammation, and mucus production (26, 31, 32, 36). Alternatively, in murine models of chronic CS exposure and emphysema, MIF prevents cell death, cellular senescence, and parenchymal tissue destruction (12, 14, 15, 33, 34). The endothelial cells of Mif-deficient mice are particularly susceptible to injury caused by hyperoxia and CS. Given that vascular dysfunction likely contributes to the pathogenesis of emphysema, these data suggest that MIF may protect against emphysema by promoting vascular homeostasis in the setting of oxidative stress (14, 16, 35).

The promoter region of the MIF gene contains a (CATT)_n tetranucleotide microsatellite repeat at position MIF-794 (rs5844672) and exists as five to eight copies (i.e., CATT_5–8) (6). In response to stimulus, ICBP90 transcription factor binding affinity is greater with increased (CATT)_n repeats. Therefore, the CATT₅ allele exhibits the lowest MIF promoter activity, while CATT_6–8 alleles have progressively higher promoter activity (6, 37). Because the polymorphism is an oligonucleotide repeat, it is not detected on commonly used single nucleotide polymorphism microarrays and therefore needs to be assayed by specialized methodology. MIF-794 CATT_5–8 polymorphisms are common (minor allele frequency >5%) and are associated with disease onset and severity in diverse pulmonary disorders, including asthma, sarcoidosis, community-acquired pneumonia, and tuberculosis (3, 7, 13, 26, 38). However, the functional significance of MIF polymorphisms in COPD has yet to be studied.

Because MIF may have paradoxical effects on important COPD-related traits, and MIF is transcriptionally regulated by a well-characterized functional polymorphism, we asked if MIF-794 CATT_5–8 alleles might contribute to COPD heterogeneity. We hypothesized that the low-expressing MIF allele (CATT₅) is associated with increased COPD severity and examined the association among the MIF-794 CATT_5–8 polymorphism, spirometry, and diffusion capacity for carbon monoxide (DL_CO) in three well-characterized cohorts designed to study patients with COPD. Two of these cohorts included subjects with human immunodeficiency virus (HIV) infection. HIV infection is an independent risk factor for COPD, regardless of CD4 count, and is associated with a younger age of COPD presentation (11). HIV infection is associated with endothelial dysfunction, which may be a mechanism that contributes to emphysema susceptibility in these subjects (20). Because the endothelial cells of Mif-deficient mice exposed to CS seem particularly susceptible to injury, we hypothesized that the low-expressing CATT₅ MIF allele may have a larger effect in HIV-infected subjects.

METHODS

We performed a cross-sectional genetic association study in three cohorts of subjects: 1) Examinations of HIV-Associated Lung Emphysema (EXHALE) cohort (4), 2) Pittsburgh cohort (18), and the 3) COPDGene cohort (30). We described baseline characteristics for each of the three cohorts using means ± SE for continuous variables and counts and percentages for categorical variables. These included sex (male/female), race (non-Hispanic white and African American), age, body mass index, smoking history (pack-years), HIV status (infected and uninfected), and measurements of COPD severity, including forced expiratory volume in 1 s (FEV₁), forced vital capacity (FVC), FEV₁-to-FVC ratio, and DL_CO. Full details have been previously described (3, 16, 27) and are summarized below.

Examinations of HIV-Associated Lung Emphysema (EXHALE) cohort.

Adult subjects with or without HIV infection were recruited from participating Veterans Affairs Medical Centers as part of a substudy of the Veterans Aging Cohort Study. Enrollment was stratified by HIV and smoking status. Written consent was obtained from all subjects, and the recruitment protocol was approved by the institutional review boards of the participating Veterans Affairs Medical Centers. Exclusion criteria included individuals with clinical diagnoses of pulmonary disease other than COPD and/or asthma or preceding respiratory tract infections within 4 wk of data collection. Hankinson reference equations were used for predicted normal values for spirometry, and Neas reference equations were used for predicted normal values for DL_CO (21, 28). DL_CO was corrected for hemoglobin.

Pittsburgh cohort.

Adult subjects with documented HIV infection and 18 yr old or greater were recruited from the University of Pittsburgh Medical Center HIV/AIDS Clinic. Written consent was obtained from all subjects and the recruitment protocol was approved by The University of Pittsburgh Institutional Review Board. Exclusion criteria included new or increasing respiratory symptoms (cough, shortness of breath, or dyspnea) or fevers within 4 wk of data collection. Crapo reference equations were used for predicted normal values for spirometry, and Miller reference equations were used for predicted normal values for DL_CO (10, 25). DL_CO was corrected for hemoglobin.

COPDGene cohort.

Adult subjects were recruited from 21 clinical study centers across the United States. Each study site obtained local institutional review boards approval to enroll subjects. The primary inclusion criteria were self-identified racial/ethnic category as either non-Hispanic whites or African-Americans between ages of 45 and 80 yr with a minimum of 10 pack-year smoking history (except nonsmoking controls). Exclusion criteria included individuals with clinical diagnoses of pulmonary disease other than COPD, active cancer under treatment, pregnant women, recent eye surgery, myocardial infarction or other cardiac hospitalizations, recent surgery, previous thoracic surgery, multiple self-described racial categories, or preceding respiratory tract infections within 4 wk of data collection. Hankinson’s reference equations were used for predicted normal values for spirometry, and Miller reference equations were used for predicted normal values for DL_CO (21, 25). DL_CO was corrected for hemoglobin, and the data were also adjusted for study site altitude. We obtained samples for sequencing the -794 CATT_5–8 MIF allele after review and approval by the COPDGene Ancillary Study Committee. Case subject selection was stratified by race and Global Initiative for Chronic Obstructive Lung Disease (GOLD) criteria. Subjects with a history of asthma and subjects without DL_CO measurements were excluded. Available samples were selected across centers.

Genotyping.

MIF-794 CATT_5–8 genotyping was performed as previously described (6). Briefly, genomic DNA was extracted from collected blood samples genotype determined by polymerase chain reaction using a forward primer (5′-TGCAGGAACCAATACCCATAGG-3′) and a TET fluorescence-labeled reverse primer (TET-laboratory: 5′-AATGGTAAACTCGGGGGAC-3′). After amplification, samples were kept at 72°C for 10 min and then 4°C before analysis. Automated capillary electrophoresis on a DNA sequencer was performed on each sample, and the CATT alleles were identified using Genotyper version 3.7 software (Applied Biosystems).

Statistical analysis.

We examined within-cohort differences in demographic variables between individuals possessing one or two CATT₅ alleles versus individuals without a CATT₅ allele, using independent samples t-test, Wilcoxon rank sum tests, and χ²-tests as appropriate. To determine if the presence of a CATT₅ allele was associated with impairment in pulmonary function, we used logistic regression models for dichotomous outcomes (FEV₁/FVC <0.70 and DL_CO <80% predicted), assuming a binomial distribution and applying a logit link function, and linear regression models for continuous outcomes (FEV₁ percent predicted and DL_CO percent predicted), assuming a normal distribution and applying an identity function. All models were adjusted for age, gender, race, body mass index, and pack-years. In the EXHALE cohort, HIV status was included in all models. We examined HIV status as a potential moderator of the associations between MIF genotype and COPD severity in the EXHALE cohort by including an HIV-by-genotype interaction term in the models. Statistical analyses were accomplished with SAS v9.4 (SAS Institute, Cary, NC).

RESULTS

Baseline demographics.

To assess the association between MIF alleles and measures of COPD disease severity, we analyzed DNA samples from three independent cohorts. We first performed analyses in the EXHALE cohort (n = 333), in which 54.1% of the subjects were HIV infected. We then utilized two confirmatory cohorts; one cohort only included HIV-infected subjects (Pittsburgh cohort, n = 122), and one cohort only included HIV-uninfected subjects (COPDGene, n = 186). The baseline characteristics of the three cohorts are presented in Table 1. In all three cohorts, subjects were predominantly male (84.4% in Pittsburgh cohort, 94.6% in EXHALE cohort, and 55.4% in COPDGene cohort). The subjects in the EXHALE cohort were predominately African-American (67.9%), while less than half of the subjects in the Pittsburgh and COPDGene cohorts were African-American (46.7 and 37.1% respectively). Participants in the Pittsburgh cohort were younger (mean age of 46.8 vs. 54.1 yr in EXHALE and 61.3 yr in COPDGene) and had less pack-years (mean of 19.3 vs. 22.5 in EXHALE and 45.3 in COPDGene). Participants in the EXHALE cohort had the lowest level of DL_CO predicted (mean level of 0.56 vs. 0.65 in the COPDGene cohort and 0.66 in the Pittsburg cohort). Other baseline demographic features and measurements of disease severity are described in Table 1.

Table 1.

Demographic, genetic, and COPD descriptive statistics for EXHALE, COPDGene, and Pittsburgh cohorts

	EXHALE	COPDGene	Pittsburgh
Total number	333	186	122
Male sex	315 (94.6%)	103 (55.4%)	103 (84.4%)
African-American race	226 (67.9%)	69 (37.1%)	57 (46.7%)
Age, yr	54.1 ± 7.8	61.3 ± 8.8	46.8 ± 8.5
BMI	28.4 ± 5.3	27.5 ± 5.4	27.1 ± 6.4
Pack-years	22.5 ± 21.9	45.3 ± 23.0	19.3 ± 20.4
HIV infected	180 (54.1%)	0 (0%)	122 (100%)
-794 CATT alleles
5–5	37 (11.3%)	19 (10.3%)	17 (13.9%)
5–6	120 (36.7%)	55 (29.6%)	46 (37.7%)
5–7	35 (10.7%)	11 (5.9%)	5 (4.1%)
5–8	4 (1.2%)	0	0
6–6	80 (24.5%)	72 (38.2%)	29 (23.8%)
6–7	40 (12.2%)	23 (12.4%)	22 (18.0%)
6–8	4 (1.2%)	2 (1.1%)	0
7–7	7 (2.1%)	3 (1.6%)	3 (2.5%)
7–8	0	1 (0.5%)	0
FEV1 percent predicted	0.91 ± 0.18 [6]	0.73 ± 0.26 [1]	0.92 ± 0.22
FEV1 <80% predicted	91 (27.3%) [6]	100 (54.1%) [1]	36 (29.5%)
FEV1/FVC	0.76 ± 0.10 [6]	0.61 ± 0.16 [1]	0.74 ± 0.11
FEV1/FVC <0.70	66 (20.2%) [6]	116 (62.7%) [1]	36 (29.5%)
DLCO percent predicted	0.56 ± 0.16	0.65 ± 0.21	0.66 ± 0.16
DLCO <80% predicted	304 (91.3%)	141 (75.8%)	103 (84.4%)

Values are means ± SD. Brackets represent missing values. COPD, chronic obstructive pulmonary disease; EXHALE, Examinations of HIV-Associated Lung Emphysema cohort; BMI, body mass index; FEV₁, forced expiratory volume in 1 s; FVC, forced vital capacity; DL_CO, diffusion capacity for carbon monoxide.

MIF genotype distribution in all three cohorts.

Subjects were then grouped based on MIF-794 (CATT)_5–8 allele microsatellites repeats. Individuals possessing one or two CATT₅ alleles (CATT_5/x) were compared with individuals without a CATT₅ allele (CATT_x/x), as previous studies have demonstrated that the CATT₅ allele has the lowest level of basal and stimulated MIF promoter activity in vitro (Table 2). The most frequent MIF genotype was CATT_5/6 in the EXHALE and Pittsburgh cohorts (36.7 and 37.7%, respectively) and CATT_6/6 in the COPDGene cohort (38.2%). Hardy-Weinberg equilibrium was not deviated from in any group (P > 0.5). Consistent with previous publications that assessed CATT allelic frequencies by race, there was an increased prevalence of CATT_5/x alleles among African-Americans in the EXHALE and COPDGene cohorts (39). Otherwise, baseline characteristics were similar between CATT_5/x and CATT_x/x groups.

Table 2.

Baseline characteristics of subjects in each cohort, based on -794 CATT allele microsatellite repeat

	CATT5/x	CATTx/x	P Value
EXHALE
Total number	196	131
Male	187 (95.4%)	123 (93.9%)	0.545
African-American	139 (70.9%)	83 (63.4%)	0.356
Age, yr	54.0 ± 7.7	54.4 ± 7.9	0.645
BMI	28.3 ± 5.2	28.4 ± 5.3	0.883
HIV infected	113 (57.7%)	62 (47.3%)	0.067
Pack-years	21.8 ± 19.6	23.3 ± 25.2	0.555
COPDGene
Total number	85	98
Male	46 (54.1%)	55 (56.1%)	0.786
African-American	39 (45.9%)	27 (27.6%)	0.010
Age, yr	60.7 ± 7.9	62.0 ± 9.5	0.338
BMI	26.8 ± 4.9	28.1 ± 5.6	0.107
HIV infected	0	0	/
Pack-years	45.4 ± 24.2	45.1 ± 22.3	0.938
Pittsburgh
Total number	68	54
Male	54 (79.4%)	49 (90.7%)	0.087
African-American	38 (55.9%)	19 (35.2%)	0.031
Age, yr	49.1 ± 8.2	48.6 ± 8.9	0.761
BMI	27.4 ± 6.58	26.7 ± 6.2	0.706
HIV infected	68 (100.0%)	54 (100.0%)	/
Pack-years	19.1 ± 22.0	20.9 ± 18.5	0.568

Values are means ± SD. BMI, body mass index. EXHALE, Examinations of HIV-Associated Lung Emphysema cohort; COPD, chronic obstructive pulmonary disease

The association between MIF genotype and COPD severity.

We then sought to determine if the MIF genotype was associated with the presence of airflow obstruction, as defined by a FEV₁/FVC ratio <0.70, or an abnormal diffusion capacity, as defined by a DL_CO <80% predicted. Using a logistic regression model, we found that CATT_5/x genotype was significantly associated with a DL_CO <80% predicted in the EXHALE cohort [odds ratio (OR): 3.41; 95% confidence interval (CI) 1.24–9.19] and in COPDGene cohort (OR: 2.21; 95% CI: 1.03–4.75) but not in the Pittsburgh cohort (Table 3). When examining the HIV CATT_5/x interaction in the EXHALE cohort, we found that HIV-uninfected CATT_5/x individuals were more likely to have a DL_CO percent predicted <80% than HIV-infected CATT_5/x subjects (OR: 9.31; 95% CI: 1.97–44.06). We did not find any associations between the CATT_5/x alleles and the presence of airflow obstruction in any cohort.

Table 3.

CATT_5/x versus CATT_x/x odds ratios and 95% confidence interval for FEV₁/FVC <0.70 and DL_CO <80% predicted

	FEV1/FVC <0.70		DLCO <80% Predicted
	Odds Ratio (95% CI)	P value	Odds Ratio (95% CI)	P Value
EXHALE	1.60 (0.80–2.98)	0.14	3.41 (1.24–9.19)	0.018*
HIV infected	2.01 (0.86–4.69)	0.11	1.13 (0.25–5.20)	0.88
HIV uninfected	1.24 (0.51–3.05)	0.64	9.31 (1.97–44.1)	0.005*
COPDGene	1.33 (0.54–3.05)	0.38	2.21 (1.03–4.75)	0.043*
Pittsburgh	0.66 (0.26–1.67)	0.38	2.55 (0.83–7.85)	0.10

Results adjusted for age, sex, race, pack-years, and body mass index (BMI). EXHALE, Examinations of HIV-Associated Lung Emphysema cohort; COPD, chronic obstructive pulmonary disease; FEV₁, forced expiratory volume in 1 s; DL_CO, diffusion capacity for carbon monoxide; FVC, forced vital capacity; CI, confidence interval.

^*Significant difference.

We then used generalized linear regression models to determine if the CATT_5/x genotype was associated with postbronchodilator FEV₁ percent predicted and DL_CO percent predicted (Table 4). In the EXHALE cohort, we found that CATT_5/x subjects had a significantly lower DL_CO percent predicted (64.36 vs. 67.82, P = 0.021). We did not identify an association between CATT_5/x genotype with DL_CO percent predicted among HIV-infected individuals, but we did identify an association between the CATT_5/x genotype and a lower DL_CO percent predicted among HIV-uninfected individuals (63.46 vs. 70.00, P = 0.023). In the COPDGene cohort, we found an association between the CATT_5/x genotype with a lower DL_CO percent predicted (60.13 vs. 65.44, P = 0.059); however, this association did not meet the prespecified cut-off point of P < 0.05 for statistical significance. We found no difference in DL_CO percent predicted between genotypes in the Pittsburgh cohort (62.22 vs. 65.59, P = 0.20). We also found no association between the MIF-794 CATT_5–8 allele and FEV₁ percent predicted in any cohort.

Table 4.

Mean value and 95% confidence interval of FEV₁ and DL_CO from linear regression models

	CATT5/x	CATTx/x	P Value
EXHALE
FEV1 percent predicted	88.32 (83.35–93.29)	88.83 (84.11–93.54)	0.81
HIV infected	90.14 (84.55–95.73)	87.65 (81.35–93.60)	0.38
HIV uninfected	87.33 (82.13–92.53)	88.83 (83.40–94.27)	0.60
DLCO percent predicted	64.36 (60.70–68.01)	67.82 (64.04–71.60)	0.021*
HIV infected	65.25 (61.00–69.50)	65.64 (60.91–70.37)	0.86
HIV uninfected	63.46 (59.41–67.50)	70.00 (65.85–74.16)	0.0023*
COPDGene
FEV1 percent predicted	70.10 (65.03–75.17)	68.25 (63.25–73.26)	0.61
DLCO percent predicted	60.13 (56.17–64.10)	65.44 (61.53–69.36)	0.059*
Pittsburgh
FEV1 percent predicted	86.82 (81.04–92.60)	82.97 (76.33–89.62)	0.28
DLCO percent predicted	62.22 (58.14–66.30)	65.59 (60.85–70.33)	0.20

Results adjusted for age, sex, race, pack-years, and body mass index. EXHALE, Examinations of HIV-Associated Lung Emphysema cohort; COPD, chronic obstructive pulmonary disease; FEV₁, forced expiratory volume in 1 s; DL_CO, diffusion capacity for carbon monoxide.

^*Significant difference.

DISCUSSION

We identified an association between a functional MIF-794 CATT_5–8 promoter polymorphism and reduced DL_CO that appears to be stronger in HIV-uninfected subjects. Among HIV-uninfected subjects, we found that CATT_5/x subjects were more likely to have a DL_CO <80% predicted and a lower DL_CO percent predicted compared with CATT_x/x subjects. This association was identified in two distinct cohorts and supports previous studies implying that MIF may protect against the development of certain COPD-related traits.

The current finding that the low-expressing CATT₅ allele is associated with decreased DL_CO suggests a causal role for inadequate MIF in the pathogenesis of COPD. Multiple cellular mechanisms have been identified via which MIF may protect against the harmful effects of chronic CS exposure. MIF promotes cell growth, wound healing, and neovascularization and inhibits oxidative stress-mediated cell injury. MIF may be important for pulmonary endothelial homeostasis, as it can activate the antioxidant transcription factor nuclear factor erythroid 2-related factor 2 in endothelial cells and upregulate the transcription of vascular endothelial growth factor (9, 24). MIF also inhibits apoptosis and cellular senescence, two pathways implicated in COPD pathogenesis. The protective effect of MIF has also been demonstrated in vivo, as genetic deletion of Mif in mice increases susceptibility to CS-mediated emphysema.

Studies of MIF in patients with COPD have also suggested that inadequate MIF may contribute to COPD pathobiology. We previously found that while current and former smokers without disease have increased plasma MIF, subjects with COPD have decreased plasma MIF. This finding has been supported by two others studies. Fallica et al. (14) demonstrated decreased serum MIF in subjects with severe COPD, and Bahr et al. (5) demonstrated that gene expression of MIF in peripheral blood mononuclear cells, as measured by microarrays, was inversely correlated with COPD severity. However, not all studies have found decreased MIF in subjects with disease. Husebo et al. (22) found increased plasma MIF in early stage disease (GOLD I and II) compared with healthy smokers, although they also demonstrated a nonsignificant trend toward decreased plasma MIF in those with more severe disease (GOLD III and IV). These studies are limited by sample size and disease heterogeneity. Additionally, MIF is acutely secreted in response to cellular stress, and therefore, measuring circulating MIF at any given time may not reflect chronic levels, further suggesting the need to study the functional CATT polymorphism.

While COPD severity is associated with both airflow obstruction and a reduced DL_CO, we did not identify an association between CATT_5/x genotype and airflow obstruction. This may be due to different pathologic processes reflected by spirometry and DL_CO. In COPD, airflow obstruction reflects the consequences of airway disease and emphysema, while DL_CO reduction reflects the consequences of emphysema and/or pulmonary vascular disease (17, 19, 27). Because COPD involves multiple cellular mechanisms, proteins may have both deleterious and protective effects depending on context. This is particularly relevant when considering a role for MIF in COPD pathogenesis; MIF promotes airway hyperresponsiveness and inflammation in murine models of asthma and bronchitis but protects against CS-induced emphysema in murine models of COPD (26, 32, 36). Therefore, CATT_5/x individuals may only be susceptible to certain pathophysiologic traits associated with COPD. Alternatively, DL_CO reduction can occur due to interstitial lung disease, pulmonary hypertension, and congestive heart failure, and further studies are necessary to understand the mechanism(s) underlying the identified association between CATT_5/x individuals and reduced DL_CO percent predicted. In the future, we plan on looking at the association of MIF alleles with radiographic emphysema and echocardiographic evidence of pulmonary hypertension.

Despite a trend toward decreased DL_CO among HIV-infected CATT_5/x subjects, this study did not identify a statistically significant association between the MIF CATT₅ allele and any indexes of COPD severity in HIV-infected subjects. This may be due to confounding factors or because the sample size is underpowered to detect a difference. However, this finding may reflect underlying differences in the pathobiology of COPD between HIV-infected and HIV-uninfected individuals. Alternatively, it might suggest that HIV infection and decreased MIF expression have the same cellular consequences in the setting of chronic CS exposure, therefore nullifying the effects of the low expressing CATT₅ allele.

The strength of this genotypic-clinical analysis includes the concordant results observed in two distinct populations, despite underlying differences in cohort characteristics. To avoid a systemic bias by choosing one method to calculate percent predicted values, we used the methods from the original studies and still found an association between the CATT_5/x genotype and reduced DL_CO in two cohorts. However, a key limitation of our finding is the uncertainty that systematic differences in MIF-794 CATT allelic frequency may exist across the spectrum of disease severity, particularly because the frequency distribution of the CATT alleles varies among racial groups (29, 39). While self-reported race was included as a covariate for the statistical models, population stratification can still occur as self-reported race does not accurately reflect genetic ancestry. Additionally, smoking status can potentially be a confounder. However, current smoking status is highly correlated with pack-years because everyone who is a never smokers has a zero pack-year history, and both current and former smokers are positively correlated with pack-years. Thus including both the estimates of the factor (pack-years and smoking status) and the variances of each factor would be biased. Since a continuous measure, such as pack-years, always has higher power to detect relationships than categorical variables (smoking status), pack-years was chosen as the covariate of choice.

In summary, we identified an association between decreased DL_CO and the low expression MIF-794 CATT₅ allele in two cohorts of HIV-uninfected subjects. This finding supports previous studies suggesting that decreased MIF may contribute to disease progression in COPD, including experimental studies in mouse models. It also raises the possibility that individuals with the CATT₅ allele may be prone to a particular subtype of COPD that involves susceptibility to the development of a reduced DL_CO. Further studies are necessary to determine if knowledge of an individual’s CATT genotype can be utilized to improve prognosis or to tailor therapy.

GRANTS

M. Sauler is supported by National Institutes of Health (NIH) Grant K08-HL-135402-01 and Flight Attendant Medical Research Institute (FAMRI) Young Clinical Scientist Award Program 142017. A. Morris is supported by NIH Grants K24-HL-123342, R01-HL-120398, R01-HL-125049, and UL1-TR-001857. R. Bucala is supported by NIH Grants 1R01-HL-130669 and 5-R01-AR-049610. K. Crothers is supported by NIH Grants R01-HL-126536 and R01-HL-090342. P. J. Lee is supported by NIH Grant R01-HL-138386; Veterans Administration, Office of Research and Developmen (VAORD) Grant 11858595; Department of Defense Grant PR150809; and FAMRI Grant 150074. C. Ramsey and H. Allore are supported by the Claude D. Pepper Older Americans Independence Center at Yale University School of Medicine (NIH Grant P30-AG-021342). The COPDGene study is supported by NIH Grants U01-HL-089897 and U01-HL-089856. The COPDGene study (NCT00608764) is also supported by the COPD Foundation through contributions made to an Industry Advisory Committee comprised of AstraZeneca, Boehringer-Ingelheim, GlaxoSmithKline, Novartis, and Sunovion.

DISCLOSURES

Yale University has applied for patent protection for the clinical utility of MIF genotype determination. R. Bucala is listed as coinventor on this application.

AUTHOR CONTRIBUTIONS

M.S. conceived and designed the research; C.Z., C.R., A.B., L.Y., Y.S.S., H.M., and M.S. performed experiments; C.Z., C.R., A.B., L.Y., L.L., K.M., H.A., R.B., and M.S. analyzed data; C.Z., C.R., K.M., M.M., H.A., and M.S. interpreted results of experiments; C.Z. and M.S. prepared figures; C.Z., C.R., and M.S. drafted manuscript; C.Z., C.R., M.M., A.M., K.C., R.B., P.J.L., and M.S. edited and revised manuscript; C.Z., C.R., A.M., K.C., R.B., P.J.L., and M.S. approved final version of manuscript.