INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Few studies (1) have investigated the chronological evolution of pulmonary hemodynamics in patients with chronic obstructive pulmonary disease (COPD) not receiving long-term oxygen therapy (LTOT) or vasodilator drugs. These studies have generally included patients with pulmonary hypertension (PH) at the onset (1). It has been observed that the progression of PH was slow with a chronological change of pulmonary artery mean pressure (Ppa) of + 0.5 to 0.6 mm Hg/yr as a mean (5). It could be of interest to know the "natural history" of pulmonary hemodynamics in patients with COPD, less severely disabled and not exhibiting PH at the onset of the follow-up. The questions one would like to answer are the following: (1) Is there a significant progression of Ppa over a relatively long period (more than 3 yr and up to 10 yr)? (2) Does this progression, if present, lead to the occurrence of resting PH in the majority of the patients at the end of the follow-up period? (3) Are the patients whose exercising Ppa is abnormally high at the onset more prone to develop resting PH with time? (4) What are the predictors of the development of resting PH in these patients? (5) May the knowledge of these predictors have practical consequences in the management of patients with COPD?

To try to answer these questions, we have investigated prospectively a large cohort of 131 patients with COPD, without PH at the onset. The time period between the first and the second right heart catheterization was at least of 3 yr and was generally  5 yr. An exercising test was performed during the initial catheterization to determine the presence or absence of an abnormally high Ppa during submaximal steady-state exercise.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

One hundred and thirty-one patients with COPD were included in this prospective study. Patients had been referred to our laboratory for a complete evaluation of COPD including measurement of pulmonary volumes, arterial blood gases, and right heart catheterization. COPD was defined according to the criteria of the American Thoracic Society (6): a history of productive cough for  3 mo during at least 2 successive years and an FEV₁/vital capacity ratio of less than 60%, the total lung capacity being more than 80% of the predicted value.

All the patients were smokers or ex-smokers with a history of smoking equivalent to at least 20 pack-years. None of the patients had a history of atopy or a significant reversibility of airflow obstruction (FEV₁ increasing by > 15% after inhalation of 400 µg of salbutamol via a metered-dose inhaler).

Patients with severe obesity (BMI > 35 kg/m²), left heart disease, known obstructive sleep apnea syndrome, other chronic pulmonary disease, or any other severe disease were excluded. No patients received long-term oxygen therapy (LTOT) or drugs susceptible of having vasodilator effects (calcium channel blockers, nitrates, angiotensin-converting enzyme inhibitors).

Investigations were never performed during an acute stage of the disease. The patients were in a stable state, at distance (minimum of 6 wk) from any exacerbation: symptoms of COPD, especially the grade of dyspnea, were unchanged, the arterial blood gases (ABG) values had not worsened as compared with the best previous values, and pH was > 7.35.

The first catheterization (T0) was performed for diagnostic purposes after an informed consent was obtained from the patients. We included in the study only patients whose resting Ppa was < 20 mm Hg (without PH) at T0. The second catheterization (T1) was performed at least 3 yr after, and generally 5-6 yr after, an informed consent being obtained from all the patients.

Methods

Conventional spirography was performed using a 10-L closed-circuit water-sealed spirograph. Static volumes were measured by the closed-circuit helium dilution method. Reference values were those of the European Respiratory Society (7).

Our technique of right heart catheterization has been described previously (4). It is a minimally invasive procedure, which is performed in our laboratory after an informed consent has been obtained from the patient. Briefly, the hemodynamic measurements were always done in the morning in the supine position, without premedication, 2 h after a light breakfast. We used small-diameter floated Grandjean (8) catheters (4F; Plastimed, Saint-Leu-La-Forêt, France). The catheter was introduced percutaneously into an antecubital or a femoral vein. The pulmonary artery pressures and particularly the mean pulmonary artery pressure (Ppa) were measured. We also measured the right atrial pressure (Pra) and whenever possible the pulmonary artery wedge pressure (Ppaw). The cardiac output was obtained according to the Fick principle applied to oxygen. Arterial blood samples were obtained during heart catheterization, the patients breathing room air. The arterial pressure of oxygen (Pa_O2), carbon dioxide (Pa_CO2), and pH were measured using a blood gas analyzer (model 280 Blood Gas System; Ciba Corning, Cergy Pontoise, France). Measurements were made at rest and during a steady-state exercise performed in the supine position. Exercise was done on a bicycle-ergometer for at least 6 min with a load of 40 W or less. Electrocardiogram and pulmonary arterial pressure were continuously monitored.The exercise test was performed in all individual patients during the first catheterization (T0). It was not systematically performed during the final catheterization (T1): it was refused by some patients and some others were too breathless to perform a steady-state exercise. Of 131 patients, 79 underwent an exercise test at T1.

The same investigations (pulmonary volumes, arterial blood gases, right heart catheterization) were performed again after a delay of 3 to 18 yr. During this interval, patients were regularly followed-up at the outpatient department with quarterly or half-year check-ups. Long-term therapy usually consisted of bronchodilator drugs (salbutamol, ipratropium bromide, long-acting theophylline) and chest physiotherapy.

Statistics

Results are expressed as means ± standard deviation (SD) or medians with interquartile range (IQR), 25th to 75th percentiles, unless specified otherwise. Comparisons between groups were performed by means of the Student's t test for unpaired or paired data, as needed. Correlations were calculated using Pearson's correlation test. The chi-quare test was performed to compare differences between two independent proportions. Comparisons of predictor variables were performed by means of linear or logistic regression analysis. Significance was set at the 5% level.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The main baseline characteristics of the patients at initial evaluation (T0) are summarized in Table 1. The majority of patients (125 of 131) were male. Age at inclusion was 54.8 ± 8.3 yr (mean ± SD). Airway obstruction was moderate to severe with a mean FEV₁ of 1.46 ± 0.55 L. According to ATS criteria for assessing the severity of COPD (6), 28.2% of the patients were in stage III (FEV₁ < 35% predicted), 45.3% were in stage II (FEV₁, 35 to 49% predicted), and 26.5% were in stage I (FEV₁  50% predicted). As could be expected in patients with COPD, there was a significant but rather loose correlation between FEV₁ % predicted and resting Ppa (r = 0.38, p = 0.001) and exercising Ppa (r = 0.28, p = 0.01). The group as a whole showed a mild hypoxemia with a mean Pa_O2 of 67.0 ± 10.4 mm Hg and no hypercapnia. By definition (inclusion criteria), no patient had resting PH, and the mean resting Ppa was normal. The exercising Ppa was abnormally high (> 30 mm Hg) in the majority (76 of 131) of the patients (mean value = 32.4 ± 7.5 mm Hg).

The 131 patients with COPD had two complete sets of investigation at T0 (first investigation) and T1 (second investigation). The mean interval between T0 and T1 was 6.8 ± 2.9 yr. According to the results of the first right heart catheterization (T0), two groups of patients were defined: group 1 (n = 55) without exercising PH (exercising Ppa  30 mm Hg) and group 2 (n = 76) with exercising PH (exercising Ppa > 30 mm Hg). Group 2 compared with group 1 showed a higher resting Ppa, and at a less significant level, a lower FEV₁ and a lower Pa_O2, as shown in Table 2. Although group 2 patients were significantly older (p = 0.001), there was no significant correlation between age and resting or exercising Ppa.

We examined which proportion of patients developed resting PH during the follow-up and whether there was a difference in this regard between the two groups. The occurrence of resting PH at T1 was of 33 of 131 patients in the group as a whole: 9 of 55 in group 1 and 24 of 76 in group 2, p = 0.048. In the patients with PH, Ppa was in fact mildly elevated, ranging from 20 to 42.5 mm Hg, with a mean of 26.8 ± 6.6 mm Hg.

We compared the evolution of resting and exercising Ppa in the two groups defined above, between the first (T0) and the second (T1) right heart catheterization. The results are presented in Table 3. Table 4 shows the evolution of other hemodynamic variables between T0 and T1 for the group as a whole. It can be seen that the evolution of resting Ppa was similar in groups 1 and 2 with a very modest (but significant) increase in both groups. For the group as a whole, the annual progression of Ppa was slow with time, with a yearly change of + 0.4 mm Hg as a mean. The yearly change was + 0.28 ± 0.75 in group 1 and + 0.48 ± 1.22 mm Hg in group 2. The difference between the yearly changes was not significant.

The comparison of exercising Ppa between T0 and T1 was limited to 79 patients, 36 from group 1 and 43 from group 2 (Table 3). There was a significant (but very slight) increase in exercising Ppa for the group as a whole, the average change being about 3 mm Hg (p = 0.004). Considering subgroups, exercising Ppa significantly increased in group 1 (p = 0.001) but not in group 2. There were no significant changes in other hemodynamic variables from T0 to T1 for the group as a whole (Table 4) as well as in subgroups.

Patients with an accelerated worsening of resting Ppa ( 0.5 mm Hg/yr) (n = 40) differed from the remainder (n = 91) by a significant worsening of exercising Ppa between T0 and T1 (+ 9.2 ± 8.5 versus + 0.1 ± 6 mm Hg, p = 0.001), whereas the changes of FEV₁ and Pa_O2 were rather similar. These patients could not be distinguished at the onset (T0): their age, pulmonary volumes, arterial blood gases, and pulmonary hemodynamic data were not statistically different from those of the remainder.

For the group as a whole, the difference of resting Ppa between T0 and T1 was significantly correlated with the difference of exercising Ppa between T0 and T1 (n = 79, r = 0.63, p = 0.001) (see Figure 1), but also with the difference of Pa_O2 between T0 and T1 (r = 0.19, p = 0.03) and with the difference of Pa_CO2 between T0 and T1 (r = 0.41, p = 0.001). There was no significant correlation between the change of resting Ppa and that of FEV₁ from T0 to T1.

View larger version (19K): [in this window] [in a new window]   Figure 1.   Scatterplot of resting and exercising Ppa changes between the first (T0) and the second (T1) right heart catheterizations.

Because patients were restudied over a wide range of times (from 3 to 18 yr, with a mean of 6.8 ± 2.9 yr), it seemed necessary to analyze the correlations between the time (in years) from T0 to T1 and the longitudinal evolution of the most important variables of this study, namely Ppa and Pa_O2. In fact, there was no significant correlation between the time interval T0-T1 and the chronological change in Pa_O2 (r = 0.12, NS) (Figure 2). Similarly, there was no correlation between this time interval and the chronological change in Ppa (r = 0.13, NS) (Figure 3).

View larger version (19K): [in this window] [in a new window]   Figure 2.   Scatterplot of PaO2 changes and time interval between the first (T0) and the second (T1) right heart catheterizations.

View larger version (20K): [in this window] [in a new window]   Figure 3.   Scatterplot of resting Ppa changes and time interval between the first (T0) and the second (T1) right heart catheterizations.

According to the results of the second catheterization (T1), two groups of patients were defined: group A (n = 98) without resting PH (Ppa < 20 mm Hg) and group B (n = 33) with resting PH (Ppa  20 mm Hg). We have compared (Table 5) the baseline characteristics of the patients at T0 depending on whether they developed resting PH at the second evaluation. Patients of group B had higher resting and exercising Ppa at T0. They had also significantly lower resting and exercising Pa_O2. Age at the time of the first right heart catheterization was similar in groups A and B. However, the time interval between T0 and T1 was significantly longer in group B patients compared with group A patients (7.8 ± 3.6 versus 6.4 ± 2.6 yr, p = 0.01).

We have compared (Table 6) the chronological evolution of Pa_O2 in the two groups. The patients who developed resting PH at T1 had a slight but significant worsening of Pa_O2 (p = 0.047) whereas Pa_O2 as a mean was stable in the remainder. The mean Pa_O2 was significantly lower both at T0 and T1 in the group of patients with COPD who exhibited resting PH at the second evaluation.

We performed multiple regression analysis to determine which variables at T0 could predict the level of Ppa at T1. FEV₁ % predicted, resting and exercising Pa_O2, and resting and exercising Ppa were included in a stepwise regression analysis. It showed that resting Ppa at T1 was independently correlated with resting and exercising Ppa at T0 (r² = 0.18, p = 0.006 and p = 0.03, respectively). Because there was no significant correlation between age and resting Ppa at T0, age was not included in the multiple regression analysis. Logistic regression analysis showed that the only independent predictors (at T0) of the subsequent development of PH (at T1) were the resting and exercising Ppa at T0 (p = 0.029 and p = 0.027, respectively). Ten variables (age, time interval between T0 and T1, arterial blood gases and pulmonary volumes, resting and exercising Ppa, at T0) were tested but only resting and exercising Ppa at T0 could be included in the equation of predictors of the presence of PH (at T1). The classificatory power of the model obtained by logistic regression analysis was low. It predicted successfully 95% of the patients without PH at T1. However, only 17% of patients with PH at T1 were correctly predicted by this model. The total correct predicted rate of the model was 76%.

We also achieved multiple regression analysis to explain the changes of Ppa between T0 and T1. Only the changes of variables, significantly correlated with the changes of resting Ppa, were included in the regression analysis (i.e. changes of resting Pa_O2, resting Pa_CO2, and exercising Ppa). Changes of pulmonary wedge pressure, which were significantly correlated in univariate analysis with changes of resting Ppa (r = 0.385, p = 0.001), were not included in the multivariate analysis because the inclusion of this variable reduced the number of valid observations to 45. The regression analysis showed that the only independent variable related to the changes of resting Ppa was the chronological change of exercising Ppa (r² = 0.34, p = 0.001). However, this analysis was limited to the 79 patients who underwent an exercise test at T1.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

First, some possible biases linked to the modalities of recruitment of the patients must be emphasized: patients had to have an initial Ppa < 20 mm Hg, to survive at least 3 yr after the first right heart catheterization and to accept a second right heart catheterization. We have observed in a previous study (5) that this way of recruiting the patients did not result in a different decline in pulmonary function when compared with that observed in a classic series from the literature (9). The patients included in the present study were collected during a large time span between 1980 and 1990. These patients were not included in our previous studies (4, 5). They belonged to a large cohort (approximately 1,000) of patients with COPD who had undergone a right heart catheterization during this 10-yr period. Interestingly, we have observed that the average resting Ppa at the onset for the group as a whole (15.2 ± 2.7 mm Hg) was very similar to that observed in our laboratory in previous studies having included subgroups of patients without resting PH (5, 12), many not having undergone a second right heart catheterization: the average Ppa ranged from 15 to 16 mm Hg. Accordingly, we believe that there were no recruitment biases.

The most important finding of the present study is that in patients with COPD, the progression of Ppa with time is rather slow, the average change for the group as a whole being + 0.4 mm Hg/yr. Furthermore, the change of Ppa was not different in patients whose exercising Ppa at T0 was abnormally high (+ 0.48 mm Hg/yr) and in those whose exercising Ppa was within normal limits (+ 0.28 mm Hg/yr). For the group as a whole, Ppa increased by not more than 3 mm Hg as a mean during a 6-yr follow-up. Indeed, this very modest progression of Ppa had been observed in earlier studies (2), as the average increase of Ppa was 0.5 to 0.6 mm Hg/yr when collecting together data from three studies (2), which concerned 163 patients with COPD; it was of 0.5 mm Hg/yr in another study (5) investigating 93 patients. However, all these series (2) have included a relatively high number of patients with PH at the onset, whereas the present study was intentionally limited to patients not exhibiting pulmonary hypertension at the onset of the follow-up. It thus appears that the long-term changes in Ppa are similar, as a mean, in patients with COPD with or without pulmonary hypertension. It must be remembered that the progression of Ppa may be very different when hypoxemia is profound, as it is in the nontreated patients of the MRC study (13) or in patients whose pulmonary hypertension is severe (Ppa > 40 mm Hg) (1).

Because of these very modest changes of Ppa over an average follow-up period of more than 5 yr, pulmonary hypertension, which was defined by a resting Ppa  20 mm Hg, was present in only a minority of patients at T1: 33 of 131 (25%). This indicates that in patients with moderately severe COPD (mean FEV₁ at the onset = 1.46 ± 0.5 L, 44.6 ± 15.7% of the predicted value) the natural progression of the disease does not necessarily lead, with time, to pulmonary hypertension. This probably explains why pulmonary hypertension is far from being the rule, even in patients with advanced COPD: in an earlier study from our group (12) including 175 patients with COPD with an average FEV₁ of 1.2 ± 0.5 L and an average Pa_O2 of 63 ± 11 mm Hg, resting PH was observed in 62 of 175 patients, that is in about one-third of the patients.

The patients whose exercising Ppa at the onset (T0) was abnormally high ( 30 mm Hg) were more prone to develop with time resting pulmonary hypertension, which was observed, at T1, in 24 of 76 patients (32%) versus only 9 of 55 patients (16%) in group 1 (p = 0.048). In fact, the longitudinal evolution of Ppa was not different in groups 1 and 2, and we have noticed above that the average yearly change of Ppa was almost identical in the two groups. Patients from group 1 differed from patients from group 2 by having a significantly lower Ppa at T0 (14.1 ± 2.8 versus 16.0 ± 2.3 mm Hg, p = 0.001). Because Ppa was lower at the onset in group 1 and because the rate of progression of Ppa was nearly the same in the two groups, it ensues that a lower percentage of group 1 patients developed pulmonary hypertension (Ppa  20 mm Hg) at T1. In good agreement with these data, the logistic regression analysis showed that resting Ppa was an independent predictor at T0 for the subsequent development of PH.

Thirty-three patients (25%, group B) developed PH at the end of the follow-up (T1). In these patients, the level of pulmonary hypertension was generally mild with a final Ppa ranging from 20 to 42.5 mm Hg, and an average value of 26.8 ± 6.6 mm Hg. This is in agreement with several earlier studies, (2, 12) that have emphasized the modest degree of pulmonary hypertension in patients with COPD. Patients from group B differed from the remainder (who had no PH at T1) by having a more severe impairment of arterial blood gases (p = 0.005) at the onset, but in fact the overlapping of individual results of arterial blood gases from groups A and B at T0 was important and did not permit the occurrence of PH in individual patients to be predicted. The compared evolution of Pa_O2 during the follow-up (Table 6) is of greater interest: Pa_O2 was stable in group A, whereas it decreased (p = 0.047) in group B from T0 to T1. Consequently, the final (T1) Pa_O2 was markedly lower in group B (60.3 ± 9.3 versus 69.7 ± 11.4 mm Hg in group A, p < 0.001). This means that the patients at risk of developing PH are those who exhibit, with time, a significant worsening of hypoxemia. These patients cannot be distinguished from the others at the onset (5). Therefore, arterial blood gases must be checked regularly in patients with advanced COPD, and the occurrence of PH can be suspected in those who exhibit progressive blood gas deterioration.

It is generally accepted that alveolar hypoxia is the most contributive factor to PH in patients with COPD exhibiting severe chronic long-standing hypoxemia (Pa_O2 < 55-60 mm Hg). Hypoxemia certainly contributed to the increase of Ppa in some of our patients. In group B (patients with PH at T1), 10 patients had a Pa_O2 < 60 mm Hg at the onset and 14 at the second evaluation. On the other hand, when we compared the evolution of pulmonary hemodynamics according to the initial level of Pa_O2 at T0, it appeared that the chronological changes of Ppa were identical in the minority of patients with initial marked hypoxemia (Pa_O2 at T0 < 60 mm Hg, n = 29) and in those with initial Pa_O2  60 mm Hg (n=101): + 0.46 ± 0.82 mm Hg/yr in the former versus + 0.35 ± 1.08 mm Hg/yr in the latter (NS). Of interest, when we compared the minority of patients who had become markedly hypoxemic (< 60 mm Hg) at T1 to the remainder, the former had an accelerated increase of Ppa with time: + 0.78 ± 1.46 versus + 0.22 ± 0.80 mm Hg/yr (p = 0.01). It thus appears that the worsening of hypoxemia during the follow-up (as discussed above) rather than the level of Pa_O2 at the onset was the major risk factor for the development of PH.

In patients with Pa_O2 > 60 mm Hg the role of hypoxemia in the development of PH is probably less important and other factors may trigger the appearance of pulmonary vascular lesions. It is presently known that factors other than hypoxemia, for example, inflammatory factors, which have not been investigated in the present study, may initiate the remodeling of the pulmonary vascular bed in patients with COPD (14, 15). This remodeling could lead with time to increased ventilation-perfusion abnormalities, hypoxemia and further progression of the pulmonary vascular lesions.

Two recent studies of our group (16,17) are in accordance with the hypothesis that a mild to moderate hypoxemia (Pa_O2 > 60 mm Hg) is not a major contributor to the development of pulmonary hypertension in patients with COPD. In the first study (16) we have observed that in patients without marked daytime hypoxemia (Pa_O2 > 60 mm Hg) the presence of pulmonary hypertension was not linked to the presence of significant nocturnal desaturation: in these patients with an average Pa_O2 of 63 ± 3 mm Hg, Ppa could not be predicted by the level of diurnal Pa_O2 nor by the mean nocturnal Sa_O2. In the most recent study (17), we could demonstrate that in patients with mild daytime hypoxemia (Pa_O2 > 60 mm Hg) and nocturnal desaturation, 2 yr of nocturnal oxygen therapy had no effect on the evolution of Ppa in comparison with control nocturnal desaturators not receiving nocturnal oxygen therapy.

In summary, the progression of Ppa with time is very slow in most patients with COPD with mild to moderate hypoxemia and, consequently, pulmonary hypertension is observed only in a minority of them, even after a relatively long (6 yr as a mean) follow-up period. Patients whose exercising Ppa at the onset is abnormally high are more prone to develop resting PH with time, but in fact the longitudinal evolution of Ppa is very similar in these patients and in the remainder. Resting and exercising Ppa are independent predictors at T0 of the subsequent development of PH. However, linear or logistic regression models built with variables collected at the date of the initial right heart catheterization are unable to predict satisfactorily in individual patients the level or the occurrence of PH at second evaluation, performed several years thereafter.