Main

CLD of preterm infants presents a spectrum of states of severity from the most dramatic forms of BPD to subclinical cases. Mild forms, which may often appear without any initial, severe respiratory disease, are common(1). Although the pathophysiology of more severe forms of CLD has been studied and disturbed lung mechanics and reduced lung volume have been revealed(24), information about the functional aspects behind mild forms of CLD is lacking.

The mixing of fresh inspired gas with the gas remaining in the lung, making up the FRC, is a process of fundamental importance for the efficiency of gas exchange. In children and adults, gas mixing is affected in different kinds of airway disease(5, 6). This is conceivable from the fact that pulmonary gas mixing is highly dependent on factors as small airway patency and inhomogeneity of distribution of mechanical properties within the lungs(7). One hypothesis behind this study was that inefficient gas mixing might contribute to lung insufficiency also in mild, chronic neonatal lung disease.

The aim of this study was to characterize the pulmonary dysfunction in spontaneously breathing infants with mild CLD by assessing gas mixing efficiency together with FRC and parameters of respiratory mechanics and comparing them with healthy preterm infants at the same PCA. A simple bedside MBNW technique was used, and moment analysis(8) was applied to express gas mixing efficiency in terms of moment ratios.

METHODS

Subjects. Fifteen preterm infants with CLD (6 boys) were studied at 35.1 ± 2.4 (mean ± SD) wk of PCA (range 30-38 wk)(Table 1). Their mean birth weight was 979 ± 238 g(620-1415 g), and mean gestational age at birth was 26.4 ± 1.3 wk(24-28 wk). Mean weight at the time of study was 2048 ± 415 g(1340-2620 g). CLD was diagnosed if infants still needed extra oxygen at 28 d of life, and if they had had respiratory signs, starting during the first postnatal days, and chest x-ray changes. Ten infants were treated by intermittent positive pressure ventilation with continuous positive end-expiratory pressure for a median time of 7.5 d (2-20 d) and a maximal inflation pressure of 18-22 cm H2O. The remaining five were treated with nasal continuous positive airway pressure without ventilation for a median time of 16 d (8-32 d). Supplemental oxygen was given for a median time of 58 d (32-79 d), beginning at the 1st wk after birth for treatment of respiratory distress syndrome (11 infants), slowly progressive lung disease, starting with little chest x-ray changes at the 1st d of life but increasing over the first weeks (three infants), or patent ductus arteriosus (one infant). Eight infants still required supplemental oxygen, 23-40%, at the time of study. As a control group, we studied 15 preterm infants (7 boys) who had no history of respiratory disease. Mean birth weight was 1438 ± 299 g(1025-2005 g) and gestational age 29.7 ± 2.1 wk (26-33 wk). They were studied at a mean PCA of 34.3 ± 2.1 wk (29-38 wk), and the mean body weight at the day of the study was 1995 ± 400 g (1260-2650 g). Parental consent was given for the study, which was approved by the ethical committee of the Medical Faculty of Göteborg University.

Table 1 Clinical data on CLD patients

Measurements . MBNW test. The MBNW has been used for assessment of gas mixing efficiency in the lungs and for calculation of FRC(9, 10). Moment analysis of the course of MBNW has received considerable attention in adults and children as an advantageous way to analyze and express the content of information in such curves in terms of gas mixing or ventilation distribution(11, 12).

The test was performed using bypass flow equipment to ensure minimal added dead space. A rigid rubber face mask (Rendell-Baker) was attached to a 1-cm long metal tube with an inner diameter of 4 mm. The probe of a fast nitrogen analyzer (Hewlitt Packard model 47302A), with a rise time of 30 ms, was mounted into the tube so that it sampled breathing gas continuously close to the mask outlet. The other end of the tube was mounted into the side of another metal tube with an inner diameter of 15 mm conveying breathing gas at a constant flow of 8 L/min. The gas could rapidly be switched to 100% oxygen by a valve at the inlet. A pneumotachograph (Fleisch no. 1), connected to a pressure transducer (Siemens Elema ENT 32), was inserted at the outlet. Equipment dead space was assessed to be less than 1 mL.

The infants were studied in the supine position during sleep with regular respiration and no body movements. No sedation was used. The mask with the bypass system attached was fitted with a water-soluble gel (Macrogol) to fill up the remaining space and to secure an airtight connection. It was gently put over the nose and mouth of the infant. During an expiration, the breathing gas was switched to 100% oxygen, and the N2 washout proceeded until the end-tidal N2 concentration was below 2%. The time required for the washout was less than 2 min in all infants. The test was repeated once in each infant. The pneumotachograph was calibrated with known gas flows before each test. The nitrogen analyzer was calibrated with room air.

When the infant is breathing through the mask the flow signal from the pneumotachograph consists of the constant bypass flow, modulated by variations caused by ventilation. The constant flow must be above the maximal inspiratory flow of the infant but also within the linear interval of the pneumotachograph. Before sampling the constant flow is subtracted from the flow signal, and consequently the sampled signal represents breathing flow.

The flow and N2 signals were sampled at a rate of 250 Hz, digitized with 12-bit accuracy, and fed into a personal computer with a software described previously(13). In converting flow and N2 signals from analog to digital form, a second order Butterworth low pass filter with a cutoff frequency of 60 Hz was used to improve the signal-to-noise ratio. The delay of the nitrogen signal, due to the gas sampling procedure, is calculated and compensated for by the software. The end-tidal N2 concentration was determined as the mean of data points immediately surrounding the maximum concentration at the end of a breath. A zero-crossing technique was applied to discriminate between inspiration and expiration. Tidal volume was obtained by electronic integration of the flow signal. The total expired N2 volume was calculated from the cumulated product of flow rate and synchronized N2 concentration from the first breath during washout and until end-tidal N2 concentration was below 1/40 of the initial one.

FRC was calculated by dividing total washed out N2 volume by the initial end-tidal N2 concentration before washout. The LCI was computed as the ventilatory volume necessary to dilute the lung N2 volume to 1/40 of its initial concentration, divided by FRC(14).

Moment analysis was used to assess gas mixing. The moment is a measure of a distribution function, in this case a nitrogen washout curve. End-tidal N2 concentrations, normalized with respect to the end-tidal N2 concentration immediately before washout, were plotted against the number of volume turnovers, and this was treated as a continuous distribution(8). From such a distribution, moments of different orders can be calculated. The rth moment (Mr) of the nitrogen washout was defined as Equation (1) where r is moment order, x is breath number, high enough to permit η to reach 8 times FRC, ηk is the dilution number, defined as the ratio of cumulated expired volume at the end of any breath k to FRC, Ak is end-tidal N2 concentration at breath k divided by the end-tidal N2 concentration of the breath immediately preceding the washout(=k0). Equipment dead space (1 mL) was subtracted from each tidal volume. No correction was made for tissue nitrogen elimination.

As can be inferred from the equation, the moment of 0th order is Equation (2) and represents approximately the area under a hypothetical continuous washout curve constructed from the observed data and with normalized end-tidal concentration on the y axis and the dilution number on the x axis. When moments of higher orders are calculated, Equation 1 shows that the higher their order the more will observations at high dilution numbers affect the result. It can be expressed so that the tail of the constructed washout curve is given more weight. As impaired gas mixing will affect the curve relatively later at a washout test, this will be emphasized in moments of higher order. A conventional way to express this effect is to calculate the moment ratios M1/M0 and M2/M0.

Mechanics. Crs and Rrs of the respiratory system were measured using the single occlusion method(15). A rigid face mask (Rendell Baker) was attached to a Fleisch pneumotachograph no. 2, connected to a differential pressure manometer (Elema Schönander EMT 32). Pressure was measured close to the mask outlet by a catheter connected to another pressure transducer (Elema Schönander EMT 34). At measurement the mask was gently applied over the face of the sleeping infant. The flow and volume signals were shown at a monitor screen. After checking that breathing was regular, the outlet of the pneumotachograph was occluded by a fingertip for at least 0.2 s at the very first part of an expiration. At least five occlusions were performed. Flow and pressure signals were sampled at 250 Hz and digitized. The calculations were made by a computer program. First volume and pressure records and a flow-volume diagram of the occluded breath was presented on screen. It was accepted if the occlusion was at least 0.2 s, the pressure described a plateau, and a distinct, linear part of the flow-volume diagram could be identified and marked by the operator. Then the software calculated the time constant of the respiratory system from the slope of the linear part, using the data points in the defined interval, as well as the difference between the volume at occlusion and the theoretical resting volume of the system. Compliance was then calculated as the ratio of this volume difference and occlusion pressure. Resistance was finally calculated as the ratio of the calculated time constant and compliance.

Statistics. Correlation coefficients and the two sample t test were calculated with standard programs. A p < 0.05 was considered significant.

RESULTS

Gas mixing. The moment ratios M1/M0 and M2/M0 were both significantly higher in CLD infants than in normal infants (Table 2). The ratios were also strongly correlated in both groups (r = 0.92 and 0.95, respectively) (Fig. 1). Also LCI was higher in CLD infants than in the healthy group. The moment ratios were both significantly correlated to LCI for the CLD infants (r = 0.53 and 0.62, respectively) (Fig. 2), but they were only weakly correlated to body weight (r = 0.33 and 0.10). As shown in Table 3, the moment ratios M1/M0 and M2/M0 were equally sensitive for the detection of CLD infants and superior to LCI in this respect.

Table 2 Lung function measurements in healthy preterm and CLD infants
Figure 1
figure 1

Relationship between M1/M0 and M2/M0 for BPD infants ()(M2/M0 = -2.6 + 5.4M1/M0; r = 0.92) and normal infants (+) (M2/M0 = -4.2 + 5.9M1/M0; r = 0.95).

Figure 2
figure 2

Individual data and regression lines between M1/M0 and M2/M0 and LCI for BPD infants(M1/M0 = 1.1 + 0.10 LCI; r = 0.53 and M2/M0 = 2.7 + 0.60 LCI; r = 0.62).

Table 3 Abnormal lung function results among the 15 infants with CLD

Lung volume. Mean FRC/kg was 26% lower in CLD infants than in the healthy group, and the difference was highly significant(Table 2). Only two CLD infants had a FRC/kg within the 95% confidence interval of the control group. One of these had no other abnormalities in lung function. All infants with high moment ratios, except one, also had a low FRC/kg (Fig. 3). Although the reverse relation was less pronounced, 10/15 CLD infants had both low lung volumes and high moment ratios. However, moment ratios were not significantly correlated to FRC (r = 0.17 for M1/M0 and r = -0.10 for M2/M0).

Figure 3
figure 3

FRC/kg vs M1/M0 and M2/M0 for BPD infants () and normal infants (+). The dashed lines represent the 95th percentile for the healthy group.

Ventilation and mechanics. Tidal volume and respiratory rate did not differ between the groups (Table 2). Crs/kg was significantly lower in CLD infants, but this difference was no longer significant when Crs was normalized with FRC (specific compliance) (Table 2, Figs. 4 and 5). Rrs showed the same distribution in both the CLD and the healthy group (Table 2, Fig. 6).

Figure 4
figure 4

Crs/kg vs M1/M0 and M2/M0 for BPD infants () and normal infants (+). The dashed lines represent the 95th percentile for the healthy group.

Figure 5
figure 5

Crs/FRC vs M1/M0 and M2/M0 for BPD infants () and normal infants (+). The dashed lines represent the 95th percentile for the healthy group.

Figure 6
figure 6

Rrs vs M1/M0 and M2/M0 for BPD infants () and normal infants (+). The dashed lines represent the 95th percentile for the healthy group.

There was no close correlation between moment ratios and Crs/kg for CLD infants (r = 0.19 and 0.33), and only a minor proportion of the CLD infants had abnormal results on both tests (Fig. 4). Negative, significant correlations were found between moment ratios and Rrs in the same group (r= -0.52 for M1/M0 and r = -0.57 for M2/M0).

No differences in lung function parameters were found between CLD infants with and without ongoing oxygen treatment at the time of the study(Table 4).

Table 4 Comparison between CLD infants with and without oxygen therapy at the time of study

DISCUSSION

Before gas mixing efficiency could be evaluated in the newborn infants in this study, some methodologic problems had to be solved. One was gas sampling and flow measurement with a minimal added dead space in order not to affect the function to be measured. The problem was solved by a constant bypass flow technique. Another problem was the method of mathematical analysis of the nitrogen washout pattern. Different indices have been proposed and used to quantify the washout test(9, 1618). Although moment ratios do not seem to have been used in neonatal research, this approach was chosen as the technique is model-free, it utilizes all data collected, and according to experiences in adults and children, it minimizes ventilatory and lung size effects on the results when the analysis is based on normalized end-tidal nitrogen concentration as a function of FRC volume turnovers(8). As can be inferred from the defining equation, moments of orders above zero are progressively more influenced by end-tidal nitrogen concentrations toward the end of the washout. The relative effect of this can be expressed as the ratio of a moment of a specified order and the moment of the 0th order, i.e. a moment ratio. It means that this mode of calculation has a potential to magnify any delay of the ventilatory system to reduce the nitrogen content of the air spaces during washout in relation to ventilation and lung size, and to present this in quantitative terms. In accordance with the practice of other groups(19, 20), the analysis was stopped at the cumulated expired volume of 8 times FRC. Some information may have been lost by not extending the analysis longer. On the other hand, the result is less affected by tissue nitrogen elimination, and it can be compared with other studies with results obtained in the same way.

Preterm infants who do not recover normally from acute neonatal pulmonary disorders form a wide clinical spectrum of CLD from severe BPD to mild forms with few or no other signs than a prolonged need for extra oxygen. Several groups have shown that FRC, compliance, and specific compliance are all low, and resistance is high in severe BPD over the first months of life(24). Gerhart et al.(21) demonstrated the same physiologic background in infants with a more moderate disease. The group investigated in the present study represents a milder form of CLD, which has been increasingly common in neonatal units(1). Although all were oxygen-dependent beyond the 28th postnatal day, and eight still at the time of study, they recovered, and most of them had only generally reduced aeration at chest x-ray without patchy changes at that age. However, almost all showed distinct abnormalities on the washout test with reduced FRC and/or impaired gas mixing when studied at about 35 wk of PCA. Compliance of the respiratory system was also low, but this reduction was explained by the reduced FRC, as specific compliance was not significantly lower than in the healthy group. Consequently, mild CLD appears functionally as a disease mainly affecting lung volume and gas mixing properties of the lungs, but not mechanics.

When our results are compared with those from studies of more severe CLD, it seems that low lung volume is a constant and fundamental phenomenon through all grades of severity of CLD, but changes in mechanic parameters are restricted to more advanced stages. Watts et al.(22) have previously reported significantly longer lung clearance delay in infants with severe BPD compared with ventilated or nonventilated infants without BPD. Our data now show that it is possible that impaired gas mixing, although rarely assessed, is also such a constant phenomenon over different states of severity in CLD. From a functional point of view insufficient gas mixing may contribute to oxygen dependency as well as to increased ventilation.

The mixing of inspired gas with residual gas in the lungs occurs mainly within airways and air spaces beyond the terminal bronchioles and is of major importance for pulmonary gas exchange(7). Many factors may influence the process, such as inhomogeneity of ventilation, created by local differences in time constants of the lung units, or by nonuniform distribution of forces of inspiration, inspiratory and expiratory flow rates and patterns, tidal volume, dead space, and airway asymmetry(7, 2325). Pulmonary gas mixing has been shown to be affected in different kinds of airway disease in adults and in children(5, 6).

Morphologic studies of infants with lethal forms of BPD have shown that this condition is characterized by alveolar hypoplasia and reduced alveolar number(26, 27). It is likely that the reduced FRC found in this study also results from impaired growth of the lungs. The process behind the impairment of gas mixing is more obscure. As gas mixing takes place mainly in peripheral airways, one might speculate that the dysfunction is caused by pathologic involvement of small airways and alveoli. There was also a high degree of concordance between low FRC and high moment ratios among the sick infants (Fig. 3). FRC, however, was not significantly correlated with any of the moment ratios, neither in the healthy group nor in the infants with CLD, which suggests that FRC and moment ratios may reflect different physiologic aspects of CLD and that low lung volume is not a principal determinator of gas mixing efficiency in CLD. It is well documented that alveolar ducts and alveoli are dilated in BPD(26, 27). Such changes of airway structure can be expected to affect gas mixing efficiency. However, processes leading to nonhomogenous ventilation may also explain the result. They include edema formation and airway muscle contraction. It is therefore possible that the gas mixing disturbance in CLD is of a more dynamic nature than the reduction of FRC and may be influenced by pharmacologic treatment. The negative, significant correlation between moment ratios and Rrs in CLD infants is puzzling. No confounding correlation to respiratory rate was found. If there is any physiologic interpretation to the finding, it must await further studies of the subject.

LCI was also higher than the 95th percentile of the healthy group in a majority of the sick infants. Although the sensitivity of this measure was lower than for moment ratios, there was a considerable concordance between LCI and the moment ratios in the CLD infants, suggesting that the moment ratios and LCI measured similar properties of the lungs.

CLD has been attributed to a more or less concerted action of tissue immaturity, oxygen toxicity, positive pressure ventilation, inflammation, and infection(28, 29). In this study the most affected variables, FRC and moment ratios, did not show significant correlations to gestational age, duration of oxygen therapy, or mechanical ventilation, suggesting that mild CLD at least has no simple dose-response relation to these factors.

In conclusion, low FRC and impaired gas mixing stand out as fundamental aspects of lung pathophysiology in CLD, regardless of severity. Our data also showed that moment ratios, as indicators of gas mixing efficiency, appear to be sensitive tools for the detection of abnormalities in preterm infants with mild CLD. The MBNW test, followed by calculation of FRC and moment analysis, may be a simple way to assess and monitor neonatal lung function in CLD, and it may prove useful when the healing process needs to be quantified and for evaluation of therapy.