The reason why patients with COPD feel subjective dyspnea is a simple question. However, answering this question is not simple, and clinicians need to understand the mechanisms responsible for dyspnea. It is widely accepted that the major limitation to exercise performance and the perception of breathlessness in COPD can be attributed to dynamic hyperinflation, although activity limitation and dyspnea in COPD is multifactorial. This has been explained by the following mechanism [1–3]. In COPD, the end-expiratory lung volume (EELV) is elevated as compared to healthy controls. During spontaneous breathing at rest in patients with expiratory flow limitations, the EELV is maintained at a level above the statically determined relaxation volume of the respiratory system. In flow-limited patients, the mechanical time-constant for lung emptying is increased in many alveolar units, but the expiratory time available during quiet breathing is often insufficient to allow the EELV to completely decline to its normal relaxation volume, and thus air trapping results. Dynamic hyperinflation occurs in flow-limited patients under the condition of increased ventilatory demand during exercise. Since the total lung capacity does not change during activity, the decrease in the IC must reflect an increase in the dynamic EELV, or the extent of dynamic hyperinflation. With the limitation of the tidal volume increase during exercise, dynamic hyperinflation results in restrictive mechanical constraints which, in the extreme, can lead to alveolar hypoventilation during exercise. In patients with COPD, breathing to higher lung volumes increases the total respiratory work, and thus potentiates the perception of breathlessness, which favors a decrease in physical activity.
The rate and magnitude of dynamic hyperinflation during exercise is generally measured in the laboratory setting by serial inspiratory capacity measurements. O'Donnell et al. reported that the exercise endurance time, Borg dyspnea ratings at the isotime near end-exercise, and IC are very reproducible indices , and that 500 micrograms of nebulized ipratropium bromide can improve the exercise endurance time by 32% on average. This improvement correlated best with the IC improvement, but not with the FVC or FEV1 improvements, and the change in the Borg dyspnea ratings at the isotime near end-exercise also correlated well with the IC improvement . An increased IC means reduced resting lung hyperinflation. Using a similar mechanism, the use of tiotropium bromide, salmeterol, or a fluticasone propionate/salmeterol combination was associated with sustained reductions in lung hyperinflation at rest and during exercise. The resultant increases in inspiratory capacity permitted a greater expansion of the tidal volume, and contributed to improvements in both exercise endurance and exertional dyspnea [4, 7, 8].
In the present study, airflow limitation may have been a more important cause of clinical dyspnea than static hyperinflation. This clearly contradicts the above mentioned hypothesis, and the results of the laboratory exercise tests that are based upon it. Why is our result different? The first issue to consider is the different dyspnea evaluation methods used. We wanted to assess overall breathlessness during daily activities (clinical dyspnea) using the BDI score in the present study, whereas the Borg dyspnea ratings at isotime exercise has been used in most laboratory studies. Dyspnea during exercise using the Borg scale may provide a different type of information regarding dyspnea than clinical dyspnea . Therefore, if the cause of COPD dyspnea is hypothesized to be dynamic hyperinflation, then it is necessary to evaluate clinical dyspnea instead of laboratory dyspnea.
Murariu et al. used a method similar to ours, and evaluated their maximal symptom-limited exercise on a cycle ergometer. Their correlation coefficients between the Wmax with the IC and FEV1 were 0.81 and 0.54, respectively, and a multiple regression model using the Wmax as the dependent variable revealed that the IC was the only significant contributor to the Wmax. They also reported that the FEV1 was not statistically significant . Their study used the Wmax as the outcome, whereas we used clinical dyspnea instead. Although the methods of their analysis were similar, their comparison between airflow limitation and static hyperinflation resulted in completely different conclusions. Therefore, using clinical dyspnea as the outcome in our study probably explains the different results.
The main reason why dynamic hyperinflation can be hypothesized to be the main cause of dyspnea is the strong correlation between dynamic hyperinflation and dyspnea. Some researchers have argued against this hypothesis, since the presence of dynamic hyperinflation is not a universal finding during exercise . We did not directly evaluate dynamic hyperinflation, but instead used the IC, which is the index for static hyperinflation. The IC may reflect dynamic hyperinflation inaccurately. Nevertheless, in the study conducted by O'Donnell et al., the correlation between the magnitude of the changes in the IC and Borg scores was strong, and they concluded that this explained why dynamic hyperinflation was causing dyspnea. However, correlations in cross-sectional studies and longitudinal studies do not necessarily match, and a statistical approach such as correlation coefficients may not resolve this issue. Airflow limitation causes dynamic hyperinflation, and hence airflow limitation, dynamic hyperinflation and dyspnea may be considered as the top of a pyramid, and it may not be necessary to consider them in a linear, causal relationship.
In the present study, airflow limitation explained only 26% of the BDI score, and airflow limitation plus the diffusing capacity explained an accumulative 41% of the BDI score. In the literature, it is thought that dyspnea measures are moderately correlated with pulmonary function, psychological function, and walking tests . For example, a simple correlation between the BDI score and FEV1 has been reported to be statistically significant, with a correlation coefficient of 0.22-0.58 [10, 19–21]. Although as per pulmonary function, the FEV1 and FVC are often evaluated for a correlation with clinical dyspnea, the correlations between the FEV1, static hyperinflation and clinical dyspnea have not been evaluated simultaneously. In addition, to our knowledge, this is the first study which proved that the diffusing capacity was a significant contributor to clinical dyspnea. This may indicate that emphysema-predominant subjects with COPD are conscious of stronger dyspnea. Our results obtained from the stepwise multiple regression analyses also indicate that there are other unmeasured factors that explain clinical dyspnea. Wijkstra et al.  reported that the transfer factor for carbon monoxide (TLCO) was strongly correlated with the six minute walking test and with the maximal work load, and that backward linear regression analysis selected the TLCO and peak esophageal pressure during a maximal semistatic maneuver as the most significant determinants for exercise performance. However, although they discussed the mechanism of correlation between the TLCO and exercise capacity, their cause-effect relationship is still unknown. Similarly, the mechanism of correlation between the diffusing capacity and clinical dyspnea is also unknown
There are also important considerations in the clinical practice setting. A common misunderstanding is that hypoxemia is causing dyspnea, and proper oxygen administration alone is enough. We want to emphasize that oxygen administration to alleviate dyspnea in COPD patients whose PaO2 is over 60 mmHg is the wrong treatment.
Since some researchers understand that COPD is a systemic disease, we should consider that other many factors possibly related to dyspnea. Since depression and anxiety are frequent in subjects with COPD, they have been investigated for their role in clinical dyspnea . Unfortunately, a psychological assessment was not included in the present study.
We measured the BNP levels to investigate whether heart failure can play a role in dyspnea in COPD patients. It has been reported that BNP can be used to differentiate heart failure from respiratory diseases, including COPD, in patients with dyspnea . Furthermore, COPD patients were reported to have higher levels of BNP as compared to controls . Although the Spearman rank correlation test revealed a significant correlation between BNP levels and dyspnea, the stepwise multiple regression analysis did not. This does not explain what elevated BNP levels in subjects with COPD mean clinically, but the magnitude of this elevation may depend on the disease severity instead of dyspnea.
Some limitations of the present study should be mentioned. Most of the issues are related to the study design. First, this study is based just on correlation analysis, which is not the best way to detect the cause of a phenomenon. Second, although stepwise multiple regression analyses were performed to compare the relative contributions between airflow limitation and static hyperinflation on clinical dyspnea, over half of the contributory factors are still unknown. Third, we analyzed the FEV1 (L), FEV1 (%pred) and FEV1/FVC as for airflow limitation. Although the FEV1 is very popular, it may be an older index of flow limitation. Other methods, including the tidal volume over the envelope in the flow-volume loop or the negative expiratory pressure during tidal breathing, should be compared against any measurements of clinical or laboratory dyspnea. The present study was also limited by the small number of participants and distinct male preponderance of the subjects. Although the latter is typically observed in subjects with COPD in Japan, generalization of these results to women with COPD may be uncertain.