INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Failure of conventional treatment in patients with acute chronic respiratory failure is characterized by development of a rapid and shallow pattern of breathing, acute hypercapnia, and respiratory acidosis (1, 2). Although mechanical ventilation through orotracheal intubation is a life-support procedure in this clinical condition, it is associated with increased risk of complications, such as barotrauma and infections (3) as well as prolonged duration of hospitalization (4). Since different clinical trials have shown that noninvasive positive pressure ventilation (NIPPV) can improve both hypercapnia and respiratory acidosis, thus preventing tracheal intubation in selected patients (4), NIPPV is being increasingly used as first-choice ventilatory support in patients with hypercapnic respiratory failure (11).

In patients with chronic obstructive pulmonary disease (COPD), NIPPV reduces the work of breathing and improves alveolar ventilation during acute exacerbations with hypercapnic respiratory failure (12, 13). The former can be further alleviated by addition of a moderate amount of positive end-expiratory pressure (PEEP) to counterbalance intrinsic PEEP (PEEP_i) (12). It has been suggested, but not shown, that NIPPV can also facilitate the recruitment of nonventilated or poorly ventilated alveolar units, thereby improving pulmonary ventilation-perfusion (A/) mismatch (13, 14). The present study examines the effects of NIPPV on A/ distribution and O ₂ and CO₂ exchange, in patients with an acute exacerbation of COPD and hypercapnic respiratory failure, to test this hypothesis.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients

Ten male COPD patients (Table 1) with hypercapnic respiratory failure (FI_O2 = 0.22 ± 0.01; Pa_O2= 44 ± 8 mm Hg; Pa_CO2 = 71 ± 13 mm Hg; pHa = 7.31 ± 0.06, at admission in the Emergency Room) were studied within 36 h of admission. Pneumonia as a cause of exacerbation, systemic diseases, and chest wall deformities were considered exclusion criteria. Prior to the study, all patients had received conventional treatment decided by the attending physicians which consisted of inhaled bronchodilators, systemic corticosteroids, antibiotics, and supplemental oxygen therapy (FI_O2 = 0.24). The study was approved by the Ethics Committee of the Hospital Clinic, and written informed consent was obtained from each patient.

Protocol

Throughout the entire protocol, each patient was studied in a semirecumbent position using a modified full-face CPAP mask (No. 9035; Vital Signs Inc., Totowa, NJ) with no changes in the ventilatory circuit (Siemens Servo 300 Ventilator; Siemens-Elema BA, Solna, Sweden) nor in FI_O2. The latter was set as close to 0.21 as allowed by the patient's clinical condition. Arterial oxygen saturation (HP Model 56S; Hewlett Packard, Andover, MA) and ECG were continuously monitored.

Measurements at each time point (baseline; 15 and 30 min of NIPPV [NIPPV-15 and NIPPV-30]; and 15 min after NIPPV withdrawal [post-NIPPV]) were always obtained with the patient under stable conditions (as judged by ventilation and pulmonary and systemic hemodynamic variables within ± 5%), in the following order: (1) simultaneous arterial and mixed venous blood and mixed expired gas sampling for inert and respiratory gases; (2) FI_O2, F_O2, F_CO2; (3) pulmonary and systemic hemodynamics; and (4) pressure at the mouth and ventilatory pattern.

At baseline, the ventilator setting (CPAP mode with PEEP = 0 cm H₂O) allowed the patient to breath spontaneously through the ventilator circuit (15). NIPPV was implemented using the pressure-support mode of the ventilator. The level of pressure-support was titrated according to the patient's tolerance, starting at 5 cm H₂O and increasing progressively up to 10 to 16 cm H₂O. Low levels of PEEP (up to 5 cm H₂O) were added, except when leaks in the face mask were detected. At post-NIPPV, pressure-support and PEEP were removed and patients breathed again spontaneously through the ventilator circuit.

Pulmonary Gas Exchange

Arterial and mixed venous blood sampling was done through an 18-gauge plastic cannula inserted into the radial artery and a 7F triple- lumen thermodilution balloon-tipped Swan-Ganz catheter (Edwards Swan-Ganz^®; Baxter Healthcare Corp., Irvine, CA), respectively. Blood samples were analyzed in duplicate for pH, PO₂, and PCO₂ measurements (IL-1306; Instrumentation Laboratories, Milan, Italy), and values were corrected to body temperature. Inspired O₂ (FI_O2) and mixed expired O₂ (F_O2) and CO₂ (F_CO2) fractions were also measured (Medical Graphics, St. Paul, MN). Bohr dead space (VD/VT) and systemic oxygen delivery (O₂) were measured, and oxygen uptake (O₂ = [Ca_O2  C_O2] · T), CO₂ output (CO₂ = [C_CO2  Ca_CO2] · T), and R (R = CO₂/O₂) were calculated using standard formulas (16). AaPO₂ (AaPO₂ = [PI_O2  (Pa_CO2/R) + (Pa_CO2 · FI_O2 · ((1  R)/R))]  Pa_O2) was calculated using the actual R.

A/ distributions were obtained using the multiple inert gas elimination technique (MIGET) (17), whose general features have been reported in detail elsewhere (18, 19). Mixed expired inert gases were collected through a 1-L-inner-volume metallic heated box inserted in the expiratory line of the ventilator (20). Shunt was defined as the percentage of blood flow perfusing unventilated alveoli (% T to A/ ratio < 0.005), low A/ regions were those with perfusion to A/ ratios between 0.005 and 0.1, high A/ regions were those with ventilation to A/ ratios between 10 and 100, and dead space was defined as the percentage of alveolar ventilation to A/ ratios > 100. The first moment of the perfusion () and ventilation () distributions was defined as the mean A/ ratio of each distribution, and their dispersion (second moment) as the standard deviation of the blood flow (log SD Q) and ventilation (log SD V) distributions on a logarithmic scale. The variable DISP R  E* is an overall index of heterogeneity of lung function and represents the combined dispersion of both blood flow and ventilation distributions. It is the root mean squared difference between retention and excretion for the gases after correcting for series dead space dilution by means of the excretion value of acetone (21). The residual sum of squares (RSS) was assessed as a quantitative estimation of the overall experimental error in the assessment of A/ distributions.

Hemodynamics and Ventilatory Measurements

Systemic arterial pressure (Psa), pulmonary artery (Ppa), pulmonary wedge (Pw), and right atrial (Pra) pressures (HP Model 56S; Hewlett Packard) and thermodilution cardiac output (T) (HP M1012 A; Hewlett Packard) were measured in each condition. Reported T values are the mean of four consecutive measurements. Pulmonary vascular resistance (PVR) was calculated as PVR = [Ppa  Pw]/T.

Mouth pressure was sampled through a side port at the face mask (Pressure Transducer Model 143PC03D; Honeywell, Freeport, IL) and processed through a microcomputer by means of an A/D converter (DT 2801; Data Translation, Marlboro, MA) using appropriate software (Global Lab v.3.0; Data Translation). Minute ventilation was measured with a calibrated Wright spirometer and the measured value was corrected to BTPS conditions (Respirometer MK8; BOC-Medical, Essex, UK).

Statistical Analysis

Results are expressed as mean ± SD. Comparisons among different ventilatory conditions were done using an analysis of variance for repeated measurements and Scheffe F-test contrast analyses. A value of p < 0.05 was accepted as the level of statistical significance.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

At baseline, mean values of minute ventilation and cardiac output were within the normal range (Table 2). Yet, a moderately rapid and shallow breathing pattern was observed. All patients exhibited pulmonary hypertension and increased pulmonary vascular resistance. Three patients were studied breathing room air (FI_O2 = 0.21) and seven required oxygen supplementation (FI_O2 = 0.22 to 0.27). Hypoxemia and hypercapnia ranged from moderate to severe values. As expected in COPD patients, A/ inequality was the main intrapulmonary determinant of hypoxemia, while shunt was 5 ± 5% (mean ± SD). Otherwise, features of A/ distributions in these patients were similar to those previously reported in acute exacerbations of COPD (22, 23). Hypoxemia was mostly caused by increased perfusion to poorly ventilated lung units (low A/ areas). Thus, the first moment of the blood flow distribution () was low and the dispersion (log SD Q) was augmented in all the patients (Table 3). Likewise, the dispersion of the ventilation distribution (log SD V) was increased in all patients. Inert gas dead space was markedly elevated but was always lower than Bohr dead space, since the latter includes areas of high but not infinite A/ ratio while the former does not.

Effects of NIPPV

Mean values of pressure support and PEEP were 12 ± 2 (range: 10 to 15) and 3 ± 2 (range: 0 to 5) cm H₂O, respectively (Table 1). After 15 min of NIPPV (NIPPV-15), we observed a significant decrease in breathing frequency (p < 0.0001) together with an increase in tidal volume (p < 0.0001) (Figure 1). Minute ventilation increased (p < 0.05), as well as alveolar ventilation (A = E · [1  inert gas dead space]), from 3.1 ± 0.8 to 3.8 ± 1.5 and 3.4 ± 0.8 L/min at baseline, NIPPV-15, and NIPPV-30, respectively (p < 0.001). The significant increase in R (p < 0.001), in turn, reflected an increased pulmonary clearance of body stores of CO₂ (due to increased alveolar ventilation) relative to pulmonary O₂ uptake, although the independent changes in both O₂ and CO₂ were too small to be significant. Note that R was not different between NIPPV-15 and NIPPV-30. Cardiac output fell approximately 1 L/min (p < 0.005), resulting in a significant fall in mean Ppa (p < 0.05), without changes in PVR. Heart rate slightly decreased (p < 0.05) (Table 2).

View larger version (20K): [in this window] [in a new window]   Figure 1.   Individual time courses of tidal volume, breathing frequency, and arterial PCO2 throughout the study. Horizontal bars represent mean values at each time point.

Arterial PO₂ increased by 7 mm Hg (p < 0.05) and Pa_CO2 decreased by 7 mm Hg (p < 0.0001) (Figure 1) with a subsequent increase in arterial pH (p < 0.0005). Accordingly, arterial oxygen saturation (Sa_O2) also increased (p < 0.01) (Table 3). It is of note, however, that AaPO₂ increased with NIPPV (p < 0.001). Systemic oxygen delivery slightly decreased (p < 0.05).

Ventilation-perfusion inequality assessed by the overall index of A/ mismatching and the amount of blood flow perfusing poorly ventilated lung areas did not show significant changes with NIPPV (Table 3). Yet, a moderate but significant increase (worsening) in the dispersion of blood flow distribution (log SD Q; p < 0.025) was observed at NIPPV-15. Mean A/ ratio of the blood flow distribution (; p < 0.05) also increased at NIPPV-15. However, dead space, calculated using either inert gases or the Bohr equation, remained unchanged during NIPPV.

As indicated in Tables 2 and 3, physiologic changes provoked by NIPPV at 15 min (except for log SD Q) were also present at 30 min (NIPPV-30) and completely returned to baseline values with 15 min after withdrawal of noninvasive ventilation (post-NIPPV).

Arterial PO₂ values predicted from inert gas data were not different from measured values (mean difference between predicted and measured Pa_O2 was 0.5 ± 6.5 mm Hg). Quality control of inert gas measurements, as assessed by the RSS (mean: 1.6 ± 1.5 ; range: 0.2 to 7.7), fulfilled the requirements of MIGET (19, 24). Similarly, agreement between cardiac output calculated from inert gas data and that measured by thermodilution (mean difference: 0.1 ± 1.2 L/min) further support that the assumptions (17) of inert gas measurements were adequately fulfilled.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In the present study, both the increase of Pa_O2 and decrease of Pa_CO2 during NIPPV appear to be related to the development of greater alveolar ventilation due to reduced respiratory frequency and increased tidal volume. This also resulted in an increased pulmonary clearance of body stores of CO₂, hence increasing R. However, contrary to our initial hypothesis, implementation of NIPPV did not result in recruitment of nonventilated and/or of poorly ventilated alveolar units, in that no changes in the amount of blood perfusing areas with both shunt and very low A/ ratio occurred. It is of note that the observed increase of dispersion of the pulmonary blood flow distribution (log SD Q) (Table 3) did not result in a significant change in the overall index of ventilation-perfusion inequality (DISP R  E*), as discussed below. Because the moderate increase in Pa_O2 with NIPPV occurred on the steepest portion of the oxyhemoglobin dissociation curve, Sa_O2 significantly increased (Table 3), hence partially offsetting the deleterious effects of the fall in cardiac output on oxygen delivery. NIPPV slightly worsened oxygen supply to tissues, but this finding is probably unimportant since P_O2 remained unchanged (Table 3).

The short-term effects of NIPPV on the breathing pattern observed in the present study have also been documented by other investigators in patients with similar clinical conditions (10, 12). Additionally, the reduction of hypercapnia and increase in arterial pH caused by NIPPV were also similar to those observed in previous studies when sequential blood gas samples were obtained (4, 9, 10, 25, 26). Surprisingly, only a few studies have assessed the effects of noninvasive ventilation at a fixed FI_O2 on arterial respiratory blood gases (27, 28). During the breathing of room air, NIPPV has been shown to moderately increase Pa_O2 (and decrease Pa_CO2 by a similar amount) with no changes in AaPO₂, the latter calculated assuming a constant respiratory exchange ratio (R) of 0.8 (27). Although the latter findings suggest no role for improvement of A/ mismatch with NIPPV, intrinsic limitations of AaPO₂ to assess pulmonary gas exchange must be taken into account (29), particularly in patients with hypercapnic respiratory failure in whom estimation of AaPO₂ using actual R has been shown to be inconclusive (30). An interesting finding during NIPPV was the increase in AaPO₂ calculated using the actual R. The increase in AaPO₂ together with the increase in log SD Q might apparently indicate that the lung was less efficient as O₂ exchanger during NIPPV. Studies in animal models have shown that short-term respiratory acidosis is associated with enhancement of hypoxic pulmonary vasoconstriction (31), and vice versa, acute respiratory alkalosis worsens A/ inequality, as assessed by an increase in log SD Q (32). It could be speculated that the fall in Pa_CO2 during NIPPV might account for the increase in log SD Q observed in the current study by mitigating hypoxic pulmonary vasoconstriction in low A/ areas. Yet, this did not happen, as the amount of blood flow to low A/ units remained unaltered. An alternative and more likely explanation, however, is that the moderate increase in log SD Q alone observed in the present study does not reflect further worsening A/ mismatch during NIPPV. It may indicate the different functional behavior of two populations of lung units with different A/ ratios (one with very low A/ units, < 0.1, about 20% of blood flow, and the other with normal units, about 80% of blood flow; Table 3). During NIPPV, the increased A/ ratio of normal units (due to the combined effects of the increase of alveolar ventilation and the reduction of cardiac output) was not mirrored by those lung units with low A/ ratios (which, in patients with advanced COPD, may reflect the underlying structural derangements of lung parenchyma). This separation between the A/ ratios of normal and low A/ regions would increase log SD Q. This explanation is further supported by the lack of changes in the overall index of A/ inequality (DISP R  E*). Under the conditions of the present study therefore, the increase in AaPO₂ observed during NIPPV can be explained by the observed increase in R. The lack of changes in AaPO₂ (37, 37, and 40 mm Hg at baseline, NIPPV-15, and NIPPV-30, respectively) calculated as the basis of a constant assumed respiratory exchange ratio of 0.80 is further consistent with this hypothesis.

NIPPV produced clear physiologic effects in the breathing pattern, respiratory blood gases, and systemic hemodynamics. However, quite surprisingly, percent dead space estimated either by the inert gases or the Bohr equation did not fall during pressure support ventilation. The relatively high instrumental dead space used to deliver NIPPV through a full-face mask may explain in part this finding, obscuring relatively small changes in percent dead space.

In the present investigation, the analysis of the interactions among different physiologic effects provoked by NIPPV should provide some practical clues to improve the delivery of this type of ventilatory support to COPD patients with acute respiratory failure. From our results, it can be concluded that the primary goal of NIPPV should be to improve the patient's breathing pattern, which can be easily monitored in the clinical setting. Excessive emphasis to achieve high pressure support values may enhance the deleterious effects of NIPPV on cardiac output, pulmonary gas exchange, and oxygen delivery. Moreover, the potential gain of increasing functional residual capacity by using high ventilatory pressures does not seem worthwhile since recruitment of nonventilated and/or poorly ventilated lung units has not been shown to play a relevant role, at least, during short-term application of NIPPV in the present study. Delivery of high pressure support may further increase the patient's discomfort and disturb his/her pattern of breathing. However, these conclusions cannot be extrapolated to the use of NIPPV in other diseases (i.e., restrictive ventilatory defect) or different clinical conditions in clinically stable patients.