Effects of noninvasive ventilation on lung hyperinflation in stable hypercapnic COPD

Noninvasive mechanical ventilation (NIMV) has been consistently shown to reduce hypercapnia and hypoxaemia in patients with a variety of chronic hypoventilatory disorders 1. By contrast, studies with intermittent NIMV in chronic obstructive pulmonary disease (COPD) patients have yielded conflicting results, but patients with severe COPD and hypercapnia appear most likely to benefit 1, 2.

The mechanisms underlying changes in arterial blood gas tensions induced by NIMV in COPD are not completely understood. It has been suggested that NIMV may reduce work of breathing, facilitating recovery from inspiratory muscle fatigue 1, 2. However, this hypothesis was not supported by the study of Elliott et al. 3, who suggested that hypercapnia is reduced due to a combination of lower loads, lung deflation and decreased central respiratory drive.

In a previous uncontrolled study, the present authors found that NIMV improved arterial blood gas tensions in chronic hypercapnic COPD subjects 4. They postulated that the main effect of NIMV was a reduction in inspiratory load, possibly secondary to a decrease of lung hyperinflation. In order to assess the above hypothesis, the authors conducted a randomised, placebo-controlled study of the physiological effects of NIMV in COPD patients with stable hypercapnic respiratory failure.

Methods

Results

Ventilation studies

Figure 1⇓ shows the changes in the pattern of breathing from spontaneous breathing to s-NIMV or NIMV in the eight patients of each group measured during the first day of the study. As expected, using a high-level inspiratory pressure setting led to significant increases in V_T and t_E (p<0.01) during NIMV. By contrast, no changes in the pattern of breathing were found in controls.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.—

Spirograms comparing the breathing pattern during spontaneous breathing (•) with the breathing pattern during a) sham noninvasive mechanical ventilation (continuous positive airway pressure, ○) and b) noninasive mechanical ventilation (bilevel positive airway pressure, ○) in eight patients from each group. Values are presented as mean±sem. V_T: tidal volume.

In the eight controls, changes in PTP_di were not significant (from 408±113 to 438±113 cm H₂O·min⁻¹), indicating that sham ventilation with the Siemens Servo ventilator was not associated with a significant increase in the inspiratory load. In contrast, in the eight patients studied, a significant reduction in PTP_di compared with spontaneous breathing, was found after institution of NIMV (from 394±99 to 203±79 cmH₂O·min⁻¹; p<0.001), indicating a 48% reduction in diaphragmatic load during assisted ventilation.

Persistent physiological effects of noninvasive mechanical ventilation

At the end of the NIMV treatment period, a slow and deep pattern of breathing was spontaneously maintained and significant improvements in arterial blood gases were found (table 2⇓). Changes in P_a,CO2 were of variable magnitude, with six patients exhibiting a reduction between 1.06–3.06 kPa (8–23 mmHg), whereas in the other 12 subjects it ranged only from 0.53–0.80 kPa (4–6 mmHg). Those patients with greater changes showed higher baseline values of P_a,CO2 (p<0.05) and higher but nonsignificant values of PEEPi,_dyn, P_I, P_I/V_T/t_I, P_I/P_I,max, suggesting that they had a more advanced disease. Subsequently, all comparisons were performed with and without those six patients, but no differences were found.

View this table:

• View inline
• View popup

Table 2—

Arterial blood gases, spirometry and lung volumes before and after 3 weeks of noninvasive mechanical ventilation (NIMV)

The NIMV group also exhibited significant reductions in FRC and RV, and a mild decrease in TLC, whereas FEV₁, FVC, and forced mid-expiratory flow increased. Changes in FEV₁ and FVC were found to be proportional, as the FEV₁/FVC ratio did not vary. A reduction in mechanical loads and respiratory centre drive was also observed (table 3⇓), in terms of E_L,dyn, R_L, PEEPi,_dyn, P_I, TTI, and P_0.1. Although P_I,max and P_di,max increased after NIMV in comparison to baseline values, the change was not significant when compared to control subjects (table 3⇓).

View this table:

• View inline
• View popup

Table 3—

Respiratory mechanics and inspiratory muscle function before and after 3 weeks of noninvasive mechanical ventilation (NIMV)

Figure 2⇓ shows the baseline and final composite maximal expiratory flow/volume curves in absolute values for the NIMV group. After NIMV, RV was reduced more than TLC, resulting in a mean increase in FVC of 259 (145 to 372) mL, isovolume expiratory flows increased, and the convexity of the flow/volume loop remained unchanged. FRC was also found to be reduced more than TLC, resulting in an increase in inspiratory capacity of 235 (177–292) mL.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.—

Baseline (▪) and final (○) maximal expiratory composite flow/volume curves in noninvasive mechanical ventilation patients. Curves were constructed by plotting maximal expiratory flows against their respective lung volumes: total lung capacity, peak expiratory flow rate, 75, 50 and 25% of forced vital capacity and residual volume. Values are presented as mean±sem.

Correlates for changes in arterial carbon dioxide tension and dynamic lung hyperinflation

Changes in P_a,CO2 were most strongly related to ΔPEEPi,_dyn (r=0.72; p<0.0001), as shown in figure 3⇓. The changes in both parameters were closely related to changes in several lung function parameters, including changes in lung volumes, such as FRC and inspiratory capacity (table 4⇓). In addition, the changes in PEEPi,_dyn after NIMV were highly related to baseline values (r=−0.88) and the regression line was significantly different compared with after s-NIMV (p<0.001), as shown in figure 4⇓.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.—

Relationship between changes (Δ) in the carbon dioxide tension in arterial blood (P_a,CO2) and dynamic intrinsic positive end-expiratory pressure (PEEPi,_dyn). •: noninvasive mechanical ventilation patients; ○: control subjects. Regression line: ΔP_a,CO2: −2.9+4.9×ΔPEEPi,_dyn (r²=0.52; p<0.0001).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.—

Relationship between changes in dynamic intrinsic positive end-expiratory pressure (ΔPEEPi,_dyn) and baseline values. Closed circles and the full regression line apply to noninvasive mechanical ventilation patients, where ΔPEEPi,_dyn=0.26−0.43×baseline PEEPi,_dyn (r²=0.77; p<0.0001). Open circles and dashed regression line represent control subjects, where ΔPEEPi,_dyn=0.38−0.09×baseline PEEPi,_dyn (r²=0.02).

View this table:

• View inline
• View popup

Table 4—

Correlation coefficients of linear regression between changes (Δ) in arterial carbon dioxide tension (P_a,CO2) and in dynamic intrinsic positive end-expiratory pressure (PEEPi) with changes in various lung function tests

Discussion

The main findings of the present study are: 1) short-term application of intermittent NIMV to stable hypercapnic COPD patients decreases lung hyperinflation, thus reducing inspiratory loads; and 2) the concomitant reduction of hypercapnia was related to the decrease in lung hyperinflation.

The results are in agreement with those of Elliott et al. 3, who found that the decrease of hypercapnia after NIMV was related to a reduction in inspiratory loads, gas trapping, and central breathing drive. Indeed, an inverse relationship was found between the increase in V_T and the reduction in mechanical loads, such as E_L,dyn, (r=−0.70; p<0.0001) and PEEPi,_dyn (r=−0.52; p<0.01). In addition, the close relationship between the changes in P_a,CO2 and ΔPEEPi,_dyn (fig. 3⇑) suggests that reduction of dynamic lung hyperinflation is an important mechanism underlying changes in arterial blood gas tensions. It must be stressed that PEEPi,_dyn, as measured in the present study, reflects the lowest regional end-expiratory alveolar pressure and most probably underestimates the threshold load resulting from dynamic hyperinflation. Interestingly, a reduction in PEEPi,_dyn that closely resembles that found in these patients was present in the study of Nava et al. 21 after 4 weeks of NIMV. Although the changes (from 2.4±1.4 to 1.9±1.8 cmH₂O) were not significant, the number of patients studied was small and beta error could have influenced their results. Indirect evidence for changes in inspiratory loads were also described by Fernandez et al. 22, who found a sustained decrease in mean oxygen consumption after only two sessions of NIMV, suggesting a reduction in the work of breathing.

Conversely, recovery from respiratory muscle fatigue appears to be an unlikely explanation for the present findings. Although a reduction of the diaphragm load was found during NIMV, in line with previous reports 8, 23, the TT_di found in the patients during spontaneous breathing was far below the critical value associated with diaphragmatic fatigue 20. In addition, neither P_I,max nor P_di,max increased significantly after NIMV. These results are in agreement with Similowski et al. 24 and Levine et al. 25, who have shown that diaphragmatic muscle function is preserved, whereas tolerance to fatigue increased in advanced COPD. In addition, Garrod et al. 7 have found favourable results using NIMV in nonhypercapnic COPD patients, where the presence of chronic inspiratory muscle fatigue is not expected. However, a note of caution is necessary, since the first two studies were apparently performed in nonhypercapnic COPD patients and in the study of Garrod et al. 7 the simultaneous application of exercise training makes it difficult to identify the isolated effects of NIMV. Accordingly, although improbable, a primary improvement of inspiratory muscle function after NIMV, with subsequent changes in shallow breathing pattern and lung hyperinflation cannot be totally discarded. For simplicity's sake, in the following discussion it is assumed that muscle fatigue did not play a role in the present study's results.

A possible explanation for the significant reduction in lung volumes after NIMV was the marked increase in expiratory time during NIMV (fig. 1⇑), which could favour the emptying of slow lung units. It is possible that a lower end-expiratory volume may be reached at the end of each ventilation session and, thereafter, be maintained by a newly adopted spontaneous pattern of breathing. Indeed, ΔPEEPi,_dyn was found to be inversely related to changes in V_T (r=−0.52; p<0.01) and t_E (r=−0.36; p<0.05).

It could be argued that lung deflation was merely the result of a reduction in airway obstruction, revealed by the significant increase in FEV₁ and maximal expiratory flows observed after NIMV in these cases. Although some lung deflation may occur in response to bronchodilator treatment in severe COPD patients 26, this is an unlikely explanation for the results of the present study considering the magnitude of deflation and the absence of significant change in lung volumes in the control group, despite of the similar medication used in both groups. An increase in FEV₁ (p=0.051 calculated from the published data) was also found by Elliott et al. 3.

Changes in FEV₁ were proportional to those of FVC, and consequently the FEV₁/FVC ratio did not change, indicating that the improved maximal expiratory flow rates reflected increased volume recruitment, i.e. RV was reduced more than TLC with a resultant increase in vital capacity (VC; fig. 2⇑). Interestingly, the increase in maximal expiratory flows associated with lung deflation, mimics the functional effects of lung volume reduction surgery 27, 28. Indeed, a flow/volume curve similar to that shown in figure 2⇑, i.e. exhibiting selective reduction of RV relative to TLC with higher isovolume maximal expiratory flows and no significant effect on the expiratory time constant, has been recently described in COPD patients submitted to lung volume reduction surgery 28. In such patients, this type of response was explained by an increased lung elastic recoil at TLC 27, 28. Changes in VC after NIMV have been previously reported by Braun and Marino 29, Ambrosino et al. 30, and Perrin et al. 31, although in the latter study did not reach statistical significance.

Another possible explanation for the decrease in lung hyperinflation is an improvement in airway resistance. This hypothesis is consistent with the close relationship found between ΔPEEPi,_dyn and ΔR_L (table 4⇑), but the reasons why airway resistance should change following NIMV are not clear and were not addressed in the present study. It could be speculated that it was secondary to persistent correction of hypoxia, considering that a primary effect of hypoxaemia is an increase of airway resistance and its correction can reduce postinspiratory inspiratory activity and dynamic lung hyperinflation 32. In addition, the improvement of alveolar ventilation might cancel out hypercapnia-induced salt and water retention 33, thus reducing airway narrowing. Recently, Burns and Gibson 34 suggested that applying positive intrathoracic pressure by NIMV could have the effect of reducing airway wall oedema. This hypothesis might also help to explain the improvement in airway resistance in these subjects and the resulting lung deflation. However, it is also possible that changes in resistance merely reflect recruitment of normal lung, previously collapsed by hyperinflated emphysematous areas, with a consequent increase in airway tethering. If the latter is true, changes in resistance should be considered more a result than a cause of lung deflation.

A pattern of slow and deep breathing was also found after NIMV in this study and the reduction in P_a,CO2 was closely related to the increased V_T, in line with Nava et al. 21. Similar changes in the pattern of breathing have been previously reported by several authors using negative 35–37 or positive pressure ventilation 3, 21, which were attributed to an improvement in inspiratory muscle strength 21, 35–37 or in chemosensitivity to carbon dioxide 3. Conversely, in the present study, the increase in V_T was significantly related to ΔPEEPi,_dyn (table 4⇑), ΔFRC (r=−0.50; p<0.005) and ΔRV (r=−0.47; p<0.005), even in the presence of a decreased drive to breathe (P_0.1) and a small and nonsignificant increase in P_I,max.

Changes in P_a,CO2 were striking and greater than those reported in recent controlled studies 6, 21. However, this was only true for a small fraction of the patients in this study (six of 18), whereas in 12 of them, the changes were within a range of 0.53–0.67 kPa (4–5 mmHg; fig. 3⇑). As pointed out by Nava et al. 21, these findings could be explained by the presence of subgroups of patients with different levels of response to NIMV application. However, the current authors did not find nonresponders in terms of P_a,CO2 as reported by Elliott et al. 3 and by Nava et al. 21, even though the severity of COPD was very similar in the three studies, suggesting that there may be another explanation for these findings. The difference could be related to the manner in which NIMV is applied. In the above-mentioned studies 3, 6, 21, NIMV was applied during sleep and using a nasal mask, and it is known that patients using this interface may receive a lower than expected support by simply opening their mouth 38. Consequently, V_Ts delivered during NIMV could have been variable along the study period in those reports. It is noteworthy that using negative pressure ventilators, where the V_T should be expected to be stable in well-adapted patients, mean reductions in P_a,CO2 of ∼1.06 kPa (8 mmHg) were found in the studies that closely matched these results 1.

The present study suggests that COPD patients with hypercapnia and dynamic lung hyperinflation may be a particular subgroup that is likely to benefit from NIMV. Other recent studies in hypercapnic patients have given conflicting results 6, 23, 36, 39, but only two of these studies employed sham ventilation 23, 39 to rule out a placebo effect 40. Renston et al. 23 reported an improvement in exercise capacity and a reduction in dyspnoea score after only 5 days of NIMV, without changes in blood gases. Longer unsupervised studies using a mean IPAP of 12 cmH₂O were unable to show improvements in blood gases and clinically relevant benefit of NIMV 39, 41. In contrast, the study of Meecham Jones et al. 6 which used an IPAP closer to that used in the present study, showed positive results. As previously mentioned, the level of IPAP appears to be important, given that higher levels are associated with significant changes in the pattern of breathing (lower respiratory rates and greater V_Ts) and reduction of P_a,CO2 8. In addition to all the previous possibilities, the consistent changes observed in this study may be due to the close supervision of ventilation at the laboratory and the duration of treatment in a group of patients selected for presenting hypercapnia.

Another difference from previous communications is that the present study was performed in awake patients, whereas home NIMV is usually prescribed at night. As the goal was to study the physiological effects of NIMV, a daytime setting was chosen, in order to control the appropriate use of ventilators and to be able to use invasive techniques to measure respiratory system mechanics and inspiratory muscle function as needed. Although previous reports using negative pressure ventilators 22, 30, 35–37 have shown favourable results with daytime application, it should be stressed that this study was not designed to define a schedule to administer NIMV. Interestingly, daytime NIMV appears to be as effective as nocturnal NIMV in improving arterial blood gases in patients with restrictive thoracic disorders and neuromuscular disease 42. In addition, the daily time duration of NIMV in this study could be considered too short, but again it should be noted that in three recent larger studies 7, 31, 41 involving 51 patients, 17 used ventilators for ∼4 h or less due to poor tolerance. Similarly, in the frequently quoted study of Meecham Jones et al. 6 the lower limit of use was 4 h. In fact, the large variation in daily hours of use observed in the published studies, including negative and positive pressure ventilators 1, clearly reveal that an appropriate schedule is not known.

One important limitation of this study is that it was not blinded for evaluators because the poor tolerance to sham-BiPAP precluded its use. In the larger study performed by Casanova et al. 41, sham-BIPAP was also discarded due to intolerance. In spite of this limitation, patients and physicians in charge of their clinical care were blinded to randomisation, thus avoiding a potential bias related to changes in their current medication. Furthermore, spirometry and lung volumes were measured by a trained nurse unaware of the purposes of the study and the randomisation of patients. Another problem is that the sham group was assigned to CPAP using a Servo Siemens ventilator, and it is known that these ventilators can impose a significant amount of work of breathing 11. A low-level pressure was used to deal with that problem, and measurements of the PTP_di confirmed that the work of breathing during ventilation sessions was not significantly increased compared with spontaneous breathing in the sham group. It was decided not to use higher levels of CPAP, because values of ∼5 cmH₂O could significantly reduce PEEPi,_dyn and unload inspiratory muscles 43, making results in the control group difficult to compare with NIMV patients.

In summary, these results show that application of noninvasive mechanical ventilation to severe stable hypercapnic chronic obstructive pulmonary disease patients can significantly improve daytime blood-gas tensions in relation to a decrease in pulmonary hyperinflation and inspiratory loads and to the adoption of a slow deep pattern of breathing. These data do not support the hypothesis that these improvements are due to a relief in chronic inspiratory muscle fatigue.