Obesity Is Associated With Altered Lung Function Independently of Physical Activity and Fitness
Measures of obesity, especially central adiposity, have been associated with reduced lung function. However, previous studies may have been affected by confounding by physical activity and fitness. This study aimed to examine the relationship among body fatness, fat distribution, and lung function, adjusted for physical activity energy expenditure (PAEE) and aerobic fitness (VO2max), in a cohort of British white adults with a family history of type 2 diabetes. A total of 320 adults (mean age 40.4 ± 6.0 years) attended for anthropometric and VO2max testing, and had ambulatory heart rate monitoring for 4 days to determine PAEE. Spirometry was used to measure forced expiratory volume in 1 s (FEV1) and forced vital capacity (FVC). The tests were repeated 12 months later, and a cross-sectional analysis using linear regression with repeated measures was performed. Measures of obesity (BMI, waist circumference (WC), fat mass (FM), body fat percentage (BF%)) were associated with lower lung function in men and women (P < 0.01), while waist-to-hip ratio (WHR) was associated with lower lung function in men only (P < 0.001). Associations remained after adjusting for age, smoking status, height, PAEE, and VO2max. The estimated difference in mean FEV1 and FVC per unit increase in the exposure measures were consistently stronger in men compared to women (P for interaction <0.001). Obesity is inversely associated with lung function in adults, but central fat distribution appears to have a stronger relationship with respiratory mechanics in men than in women. These associations were independent of the degree of physical activity and aerobic fitness in this cohort.
Chronic respiratory disease is associated with increased mortality (1,2), but even mild perturbations in lung function, which may not be clinically apparent, have been shown to predict respiratory and all-cause mortality (3,4). Several longitudinal studies have reported an association between impaired lung function and subsequent risk of coronary and cerebrovascular disease (5,6), increased insulin resistance, and diabetes (7,8). Respiratory complications associated with severe obesity have been consistently reported, including asthma and obstructive sleep apnea syndrome (9), but there is greater uncertainty about the association between lesser degrees of obesity and lung function.
Cross-sectional studies from several different populations (10,11,12,13) have suggested such an association. However, only some of these studies have included measurements of central adiposity, and where these have been taken, they tend to correlate with worse lung function, even in nonobese individuals (12,14,15). Thus, the distribution of body fat may be an important determinant of lung function, and this may account for the more pronounced association that central adiposity has with lung function in men as compared to women (16,17). The possibility of a different strength of association in men and women is supported by longitudinal data showing that the magnitude of decreased lung function associated with weight gain tends to be more pronounced in men (18).
Some previous studies of the associations of adiposity and lung function have only used BMI as a marker of obesity, without taking body composition and fat distribution into account. Many have not considered potential confounding by physical activity, or when it has been considered, adjustment has been made using relatively imprecise methods such as questionnaires, leaving open the possibility of residual confounding. Therefore, in this study we sought to examine the relationship between the degree of body fatness and lung function in a population-based cohort of British white adults. These individuals were at increased risk of developing type 2 diabetes by virtue of having a parental family history of the condition. In particular, we precisely quantified objectively measured physical activity energy expenditure (PAEE) and aerobic fitness (VO2max) to describe their potential modifying effects on the association between obesity and lung function.
Methods and Procedures
This study utilizes data from participants in the ProActive trial, a randomized controlled trial investigating the efficacy of a family-based lifestyle intervention to increase physical activity among adults with a parental family history of type 2 diabetes. Full details of participant recruitment, study design, and measures have been previously reported (19). Complete anthropometric, spirometry, and PAEE measurements were available on 345 of the 365 participants at baseline. Measurements were repeated 12 months later as part of the ongoing trial. For the purposes of this article, these two sets of measurements were both used in a cross-sectional analysis. Ethical approval was obtained from the Eastern England Medical Research Ethics Committee, and West Suffolk, Cambridge, Huntingdon, and West Essex local research ethics committees. All participants gave written informed consent.
Anthropometry and body composition
Participants attended the testing site in the morning after an overnight fast. Weight in light indoor clothing was measured using a Seca scale and height using a rigid stadiometer. Waist circumference (WC) was measured in duplicate using a metal tape and according to a written protocol. Body composition was measured using an electrical impedance device (Bodystat, Isle of Man, UK). These data were used to calculate measures of obesity and composition including BMI, waist-to-hip ratio (WHR), fat mass (FM), fat-free mass (FFM), and body fat percentage (BF%).
A submaximal, graded treadmill test was used to estimate VO2max. This was done by plotting a regression line established between O2 uptake and heart rate during submaximal testing, and then extrapolating this to the estimated maximal heart rate (220 − age). O2 uptake and CO2 production were continuously measured by indirect calorimetry throughout the test (Vista XT metabolic system; Vacumed, Ventura, CA). The volume of expired air was measured with a turbine flow meter, CO2 concentration (FECO2) with an infrared sensor, and expired O2 concentration (FEO2) with a fast differential paramagnetic sensor. These variables were continuously analyzed using Vista Turbofit software (Vacumed, Ventura, CA) to quantify O2 consumption and CO2 production.
Resting energy expenditure was calculated by the Weir formula (20) using VO2 and CO2 measurements obtained in the fasting state, after 10 min of supine rest. Resting energy expenditure was measured for at least 6 min by a securely fitted mask. PAEE was quantified using the flex heart rate (21,22). Participants wore a heart rate monitor (Polar Electro, Kemple, Finland) continuously during waking hours for 4 days. During the submaximal exercise test, the highest resting heart rate and the lowest exercising heart rate were noted. The average of these two rates was taken to be the index rate, or “flex rate.” This rate then defined the threshold in the analysis of free-living heart rate data, to define “rest” and “exercise.” During periods when the recorded heart rate was below the threshold flex heart rate, energy expenditure was estimated to be equivalent to resting energy expenditure. During periods where the recorded heart rate exceeded the threshold flex heart rate, energy expenditure was calculated from the individual heart rate/energy expenditure regression line obtained during the exercise testing. PAEE was calculated by subtracting resting energy expenditure from the estimated average daily energy expenditure over 4 days. Aerobic fitness and PAEE were expressed relative to FFM (kJ·kg FFM/min) to adjust for interindividual differences in body size (23).
Measurements of forced expiratory volume in 1 s (FEV1) and forced vital capacity (FVC) were obtained using a Vitalograph spirometer. Each participant completed the test while sitting down, and completed two spirometry attempts. Where these two spirometry readings differed significantly (i.e., where the difference between the two FVC measurements was >100 ml), a third measurement was taken. The tests were conducted by specially trained technicians and in accordance with American Thoracic Society guidelines (24).
Descriptive summary statistics were calculated separately for men and women using means and standard deviations. We used a linear regression model with repeated measures to explore the associations between each of the obesity measures (BMI, WC, WHR, FM, and BF%) as exposures, and measures of lung function (FEV1, FVC, and FEV1:FVC ratio) as outcomes. Lack of independence among observations on the same participant was accounted for by allowing the intercepts to vary randomly among participants. We adjusted for age, sex, height, and smoking status (never, former smoker, or current smoker) in each model, except where BMI was the exposure in which case height was not included, to avoid potential colinearity. PAEE and VO2max (both expressed per unit of FFM) were introduced into the models (separately and then together) to assess whether the associations between obesity and lung function were independent of physical activity and aerobic fitness. The direction and magnitude of the associations were similar for the models examining the relationships between body fatness and lung function when adjusted for PAEE and VO2max as separate exposures. Therefore, we present only two models; model one is adjusted as described above, and model two is adjusted as per model one with the addition of PAEE and VO2max. To assess whether the effect of each obesity exposure was different in men and women, we repeated the above analysis including a sex × obesity interaction term in the models. Where the P value for the interaction term was <0.1, we report the results separately for men and women.
To check our results were not biased by nonrandom missing data, we conducted sensitivity analyses and found no difference between those with (n = 345) and without (n = 20) baseline lung function measures. We also excluded those who reported using respiratory medications at any of the time points (baseline n = 25). Complete data was, therefore, available for 320 people at baseline. Variability in the number of observations included in the analyses is due to incomplete data sets and analysis that includes baseline and follow-up data. Analyses were performed using Stata version 9.1 (STATA, College Station, TX).
The baseline characteristics of participants are shown in Table 1. There were no differences in age or BMI between genders, though men tended to be taller and heavier than women with greater WC, WHR, and lower BF%, as expected. Overall, 32% of participants had normal weight (BMI <25 kg/m2), 40% were overweight (BMI 25–30 kg/m2), and an additional 28% were obese (BMI >30 kg/m2). Men were also fitter and more physically active than women even when normalizing for FFM (Table 1) and had better lung function measures on spirometry. Current cigarette smoking was reported in 23% of men and 17% of women at their baseline visit.
Table 1. Characteristics and lung function measures of participants at their baseline visit, overall and separately for women and men
Adjusted associations derived from linear regression modeling among measures of obesity, body fat distribution, and lung function are shown in Table 2. In these analyses we observed significant inverse relationships between body fatness (BMI, FM, BF%) and lung function (FEV1, FVC). These associations remained significant after adjusting for confounders such as age, height, smoking, and gender, and also after adjusting for PAEE and aerobic fitness. Stratification by smoking status did not significantly alter the pattern of associations seen (data not shown). WC was only weakly associated with FEV1 and FVC. However, increased WHR was strongly associated with a reduced FEV1 and FVC across the whole group. There was no association between the FEV1-to-FVC ratio and any of the exposure measures after adjusting for physical activity and aerobic fitness.
Table 2. Relationship between measures of body fatness and body fat distribution with measures of lung function across the entire study sample
Tables 3 and 4 show the associations between measures of obesity and lung function, in men and women separately. Gender significantly modified the associations among WC (P for interaction <0.001), WHR (P for interaction <0.001), FM (P for interaction <0.001), and BF% (P for interaction <0.001) with FEV1, but did not affect the association between BMI and FEV1 (P for interaction = 0.27). In women, there were inverse associations among WC, FM, BF%, and FEV1 but not WHR. In men, there were inverse relations between FEV1 and all measures of obesity.
Table 3. Relationship between measures of body fatness and body fat distribution with measures of lung function in women
Table 4. Relationship between measures of body fatness and body fat distribution with measures of lung function in men
The associations between body fat measures and lung function were stronger in men than in women, as indicated by the regression coefficients, such that a 1 cm increase in WC was associated a 4 ml lower FEV1 in women and 17 ml in men. The magnitude of the associations remained similar even after adjustment for aerobic fitness and physical activity. Similarly, gender also modified the associations among BMI (P for interaction <0.0.04), WC (P for interaction <0.001), WHR (P for interaction <0.001), FM (P for interaction <0.001), and BF% (P for interaction <0.001) with FVC. In women, there were significant inverse associations between BMI, FM, and BF% and FVC but not with WC or WHR. In men, all obesity measures were significantly inversely associated with FVC. As with FEV1, these associations were independent of PAEE and VO2max in men and women. Across the whole group, the association between BMI and FVC was such that a 1 kg/m2 increase in BMI was associated with lower FVC by 266 ml in men and 88 ml in women. There were no interactions between the FEV1-to-FVC ratio and any of the obesity measures.
Our results suggest that in a cohort of British white adults with a family history of type 2 diabetes, the degree of body fatness is inversely related to lung function. This confirms findings in several other population-based studies. However, in many of these studies, the only measure of obesity reported was BMI (18,25,26). We have utilized electrical impedance measures of FM and BF% to more precisely describe the relationship between adiposity and altered respiratory function. In addition, by measuring WC and WHR, we were able to examine how the distribution of body fat affects lung function. Our findings confirm the associations between obesity and lung function and suggest that abdominal adiposity may be particularly detrimental in this regard.
Most previous studies, which have examined the association between obesity and lung function, have not taken into account the confounding effects of physical activity levels and aerobic fitness. Where these exposures have been reported in studies, they are often measured subjectively (10,13,14,17). Ours is the first report on the association between obesity and lung function to quantify these two important confounding exposures objectively, and to confirm that the association exists independently of them.
There are several potential mechanisms by which excess body fat might lead to reduced lung function, broadly categorized into mechanical and inflammatory. With increasing obesity, intra-abdominal fat deposition and accumulation may impede the descent of the diaphragm during inspiration, which would affect several spirometric variables (27). Also, increased abdominal fat has been shown to reduce the expiratory reserve volume, by displacing the diaphragm upward and reducing functional volume in the thoracic cavity (12). In addition, the deposition of fat on the chest wall may impede expansion and excursion of the rib cage, through a direct loading effect or by altering intercostal muscle function (28). A recent laboratory-based study reported that during exercise, obese volunteers show a decrease in inspiratory muscle activity as a result of reduced inspiratory strength and increased ventilatory drive (29). Furthermore, increased body fat has been shown to be associated with markers of systemic and vascular inflammation such as C-reactive protein and the hormone leptin (30,31). These inflammatory factors may exert local effects in lung tissue, leading to subtle reductions in airway diameter. However, this might be expected to lead to a reduction in the FEV1-to-FVC ratio with increasing obesity, which we did not observe in this cohort.
In our analysis stratified by gender, the inverse association between WHR and lung function was only apparent in men but not in women. This is consistent with findings from several other studies, which have confirmed an association between body fatness, but not central fat distribution and altered lung function, in women (12,16,17,32). Where studies have shown such an association, it tends to be less strong in women than in men (15). While gender differences in patterns of fat distribution would account for the stronger association between obesity and impaired lung function seen in men, by measuring WHR and WC, we have attempted to correct for this. However, the higher WHR in men compared to women in this and other studies might account for these differences, if there was a threshold below which WHR did not affect lung function, as suggested by Harik-Khan (16). For any given distribution and volume of body fat, the effects on respiratory mechanics might be more pronounced in men than in women. It may be that the absolute volume of visceral fat, rather than relative proportions, dictate the degree to which diaphragmatic movement and chest-wall expansion are reduced. Certainly, the relationships among obesity, body composition, and lung function, and the underlying mechanisms, are complex and require further exploration. More precise measurements of fat distribution, such as radiological imaging and further measures of lung volumes and function (e.g., expiratory reserve volume, gas transfer rates) would help to elucidate these associations further.
Aerobic fitness and self-reported physical activity levels have also been shown to be positively correlated with lung function (25,33,34), hence the importance of accurately quantifying and adjusting for these confounders. We also adjusted for smoking in this study, and results were consistent among smokers, former smokers, and those who never smoked. By excluding individuals who were taking respiratory medications at the time of assessment, we minimized the confounding effect of prevalent, diagnosed common lung diseases such as chronic obstructive pulmonary disease and asthma. Thus, the effects we have observed in this cross-sectional analysis are more likely to be attributable to obesity and body fat distribution than to any other potential confounder or determinant.
Longitudinal studies of lung function suggest that increased body fatness is predictive of lung function decline and that weight loss can improve both FEV1 and FVC (35,36). Decreasing body fatness might, therefore, have clinical implications in terms of improving lung function and reducing the prevalence of respiratory disorders such as sleep apnea in overweight and obese individuals. Our study has shown that the associations between obesity and lung function are independent of fitness and physical activity; therefore, improvements in lung function through a reduction in obesity could potentially be achieved with or without alterations to fitness and physical activity, through dietary mechanisms, for example. Longitudinal studies are required to further examine the interactions and effects of combined change in obesity, fitness, and physical activity with improvement in lung function.
A number of limitations should be considered when interpreting findings from this study. First, the analysis was cross-sectional which makes the inference of causality difficult. It remains to be seen whether reducing obesity and improving body composition measures will improve lung function in this cohort. Although we controlled for several potential confounders such as gender, age, height, smoking status, and prevalent respiratory disease, we cannot rule out residual confounding due to measurement error. Our results pertain to a specific cohort of healthy British white adults with a family history of type 2 diabetes. However, with the rising prevalence of obesity and type 2 diabetes, it is likely that over time our results will increasingly apply to a higher proportion of the general population.
This study also has several important strengths. Although there is significant interindividual variation in the correlation between heart rate and physical activity intensity, particularly at low-intensity levels, our use of individually calibrated heart rate monitoring will have reduced this variation to a large extent. This approach has been extensively validated, both in the laboratory setting and in free-living conditions (37,38,39). Measurement of O2 uptake during a submaximal exercise test allows the accurate prediction of VO2max. This test is safer and more feasible in larger population-based studies than a true maximal test and is unlikely to have introduced significant bias, given that the error in predicted maximal heart rate is likely to be random across the entire cohort. The analytical approach that we used improved the statistical power of the regression models, by including both baseline and follow-up measures.
In summary, we found negative associations among the degree of obesity, fat distribution, and lung function in a cohort of British white adults with a family history of type 2 diabetes. These associations were independent of objectively measured physical activity and aerobic fitness. The associations between obesity and lung function also differed between men and women with the observed effect being stronger in men. Further studies are required to elucidate the mechanisms by which obesity influences lung function, particularly in relation to the differential effects observed between men and women.
The Medical Research Council (ref no. ISRCTN 61323766), NHS R&D, RCGP Scientific Foundation, and Diabetes UK (ref no. RG35259) funded the development and execution of the ProActive trial. We thank all of the practice teams for their hard work in helping to recruit the participants and the trial coordination, measurement, and intervention teams for caring for participants at various stages of the study. We also thank the participants for their time and effort. Also thanks to Stephen Sharp for statistical advice. The ProActive research team includes Kate Williams, Julie Grant, A Toby Prevost, David Spiegelhalter (principal investigator), S.J.G. (principal investigator), Nick Wareham (principal investigator), Wendy Hardeman (principal investigator), U.E. (principal investigator), Stephen Sutton (principal investigator), and Ann Louise Kinmonth (principal investigator).
The authors declared no conflict of interest.