INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Lung function testing is a well established clinical and research tool in cooperative subjects. Maximal expiratory flow-volume (MEFV) curves are useful in adults and older children because of the phenomenon of flow limitation. Isovolume pressure-flow curves (IVPFC) have been used to demonstrate the presence of flow limitation over the majority of the MEFV curve (1). This is inferred from the independence (or dissociation) of expiratory flow from driving pressure (pressure independence). In addition, extensive studies have allowed a good understanding of the physiologic basis of these curves (2) and firm recommendations about how they should be obtained (3).

Forced expiration using a compression jacket has been used extensively to measure lung function in the VT range in infants (4). Controversy exists about whether flow limitation occurs with this method, particularly in normal infants (5). The raised-volume forced-expiration technique (RVFET) was developed as an improvement on the forced expiration technique with a compression jacket, and measures lung function from a higher starting volume (6). Minimal data exist to indicate whether pressure independence occurs with the new technique (6). To allow comparisons to be made between measurements from different laboratories, such methodologic aspects of new techniques need to be standardized (7).

The primary aim of our study was to determine whether pressure independence is reached with the RVFET. Our secondary aim was to delineate the effect of compression pressure on the flow and volume variables measured with this technique. Third, because the construction of IVPFCs is impractical in routine studies, we wished to study the possibility of using a standard compression pressure.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

Twenty-one infants (nine female and 12 male), of median age 10.5 mo (range: 1.5 to 22 mo), were studied (Table 1). Eleven infants were healthy and free from any previous or current respiratory illness. The remainder had a history of recurrent wheezing (three or more episodes or one lasting > 4 wk), with two of these infants having current symptoms. There were no significant differences between groups with respect to age, weight, or length. Although there was an excess of males in the wheezy group, this did not reach statistical significance. All infants were studied in the supine position and asleep, following an oral dose of chloral hydrate (80 to 100 mg · kg¹). All parents gave written informed consent for the study and were present during the tests. The study was performed with the approval of the Princess Margaret Hospital's human ethics committee.

Equipment

Measurements were made using the RVFET (6). With this method, prior to forced expiration, the infant's lung volume is raised above the tidal range, using a pump. We used a fan pump (Inflate-all; Coleman, Inc., Wichita, KS), which was able to produce higher inspiratory flows than the diaphragm pump used in the original study (6). The pump draws air through a pneumotachograph (screen: 0 to 100 L · min¹; Hans Rudolph Inc., Kansas City, MO) and into the infant via a soft plastic face mask placed over the mouth and nose, forming an airtight seal (King Systems, Noblesville, IN). The differential pressure across the pneumotachograph was measured with a 0 to 7 cm H₂O differential pressure transducer (PX170-07DV; Omega International, Stamford, CT) and amplified (SC14C; RHT-INFODAT, Montreal, Canada). A series of solenoid valves (Ascomation, Sydney, Australia) allowed control of inflation of the infant's lungs and forced exhalation. The valve producing the airway occlusions was a balloon valve (model 9340; Hans Rudolph). The sequence of valves was controlled by BRATLAB software (RHT-INFODAT) on a computer. A pressure transducer (FPM-02PG; Fujikura, Tokyo, Japan) measured airway opening pressure (Pao) through a port on the mask. Unlike the original study, in which Pao was limited with an underwater blowoff valve, we used integrated software to sense Pao, halt inflation, and initiate forced expiration. Inflation pressure (Pi) was preset to 20 or 30 cm H₂O. Three raised-volume inspirations were produced prior to forced expiration, to decrease the likelihood of voluntary respiratory efforts from the infant, as suggested by Castile and colleagues (8). All signals were collected and analyzed on computer with LABDAT-ANADAT 5.2 data-acquisition and analysis software (RHT-INFODAT). The effective dead space of the system was 62.8 ml, and the expiratory resistance was 0.074 kPa · L¹ · s. The system was linear for flows up to 150 L · min¹, and had a flat frequency response up to 20 Hz.

Forced expiration was achieved with an inflatable plastic jacket. This was connected to a large (60 L) pressure reservoir via a series of large-bore (> 2.5 cm) solenoid valves and tubing. After the Pao reached the preset level, rapid jacket inflation was triggered, and the expiratory valve opened after a preset time delay that allowed the jacket to fully inflate.

Progressively increasing Pj values (0 to 80 cm H₂O) were used to force exhalation. Pj was measured with a pressure transducer (FPM-02PG, Fujikura). The compression pressure provided by the jacket was measured at the airway opening by inflating the jacket against an occluded airway, as previously validated (5). The pressure trace obtained is shown in Figure 1. The pressure plateau prior to jacket inflation represents the inflation pressure (Pi). Jacket inflation leads to a secondary increase in pressure to a further plateau. The difference between the two plateaus is the transmitted pressure (Ptrans) (5). Ptrans was calculated for each Pj with a 0.5 s occlusion.

View larger version (11K): [in this window] [in a new window]   Figure 1.   Trace of airway opening pressure (Pao) against time, demonstrating the calculation of inflation pressure (Pi) and transmitted pressure (Ptrans).

FEF was integrated and volume-time and flow-volume curves produced. For each technically acceptable raised volume maneuver, we calculated FEV_0.5, FEV_0.75, and FEV₁ from the forced expiratory volume-time curve (Figure 2). In addition, from the forced expiratory flow-volume curve, we calculated FEF at 25, 50, and 75% of expiratory volume (FEF_25%, FEF_50%, FEF_75%), and forced expiratory vital capacity (FVC).

View larger version (12K): [in this window] [in a new window]   Figure 2.   Volume-time tracing of a raised-volume forced-expiratory maneuver.

Analysis

Pressure-flow curves were constructed for each infant at 25, 50, and 75% of exhaled volume, and were inspected for evidence of pressure independence. Curves of FEV_t versus Ptrans were also constructed and examined for pressure-independent behavior of the variables. This was done for curves generated with Pi values of 20 and 30 cm H₂O.

Statistical data analysis was done with multiple linear regression, but with a variance/covariance structure that reflected correlation among observational units and heteroscedasticity (non-constant error variance). In particular, terms were included in the variance/covariance structure to account for correlations within children (and hence difference between children), correlations at the same Ptrans value (similarity in measurements taken at the same pressure), and increasing error variance with increasing values of Ptrans.

The model related the outcome variables (FEV_0.5, FEV_0.75, FEF_25%, FEF_50%, FEF_75%, and FVC) to the Ptrans, with particular emphasis on whether, above a certain Ptrans, there was no further increase in the value of the variable. Regression lines were fitted to estimate the effect of Ptrans, on the outcome variables. For each curve, two lines were fitted (i.e., two intercepts and two gradients were estimated), one over the "early" (< 20 cm H₂O) and one over the "late" (> 20 cm H₂O) portion of the curve. The cutoff value of 20 cm H₂O between early and late portions of the curve was selected on the basis of qualitative examination of the pressure-flow curves. A model using 15 cm H₂O as the cutoff value was evaluated, but was found to add little to the analysis.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

As a set Pi of 20 cm H₂O, there were sufficient data from 19 subjects, for analysis, and at 30 cm H₂O there were sufficient data for analysis from eight subjects. We collected fewer data using the higher Pi, since the infants were more likely to awaken and leaks were more common. In addition, the order of collection was nonrandom, and therefore the excess of infants who had measurements made at 30 cm H₂O following 20 cm H₂O may have contributed to the lack of data, owing to changes in sleep state. The Pi measured at the mouth was 19.6 ± 0.4 (mean SEM) cm H₂O in the first group, and 29.3 ± 1.3 cm H₂O in the second. There was minimal variation in the measured Pi within an individual, with a mean SD of 0.6 (range: 0.2 to 1.6) at a set Pi of 20 cm H₂O.

Pj values ranged from 0 to 77.2 cm H₂O. This resulted in a range of Ptrans values from 0 to 41.9 cm H₂O. The proportion of pressure transmitted to the airway from the jacket (Ptrans/ Pj) was dependent on the Pj selected. There was a lower percentage transmission at low Pj values, as shown in Figure 3. The mean transmission of pressure ranged from 25% at low (< 20 cm H₂O) to 65% at high (> 40 cm H₂O) jacket pressures. Neither Pi nor health status affected the proportion of pressure transmitted.

View larger version (13K): [in this window] [in a new window]   Figure 3.   Relationship between proportion of jacket pressure (Pj) transmitted (Ptrans/Pj) to the infant's airway opening during an occlusion and the absolute pressure in the jacket. A second-order polynomial with 95% CI has been fitted. There is improved transmission at higher Pj values.

The FVC measured at Pi of 20 cm H₂O was 204 ± 63 ml (mean ± SD), or 21.5 ml · kg¹. The corresponding values for a Pi of 30 cm H₂O were 274 ± 110 ml, or 28.5 ml · kg¹.

From forced expiratory flow-volume curves generated with Pi of 20 cm H₂O, FEF was calculated at 25, 50, and 75% of FVC. Flow was then plotted against Ptrans for each of these lung volumes, and was visually inspected for evidence of pressure independence. Flow limitation was inferred if there was a portion of the curve in which flow at a given volume was independent of driving pressure (pressure independence). Pressure independence was more likely to occur at low lung volumes. It occurred in none of 14 infants at 25%, 12 of 16 infants at 50%, and 14 of 16 infants at 75% of exhaled FVC. There were equal numbers of healthy and wheezy infants in the groups that failed to show evidence of pressure independence. All curves that demonstrated evidence of pressure independence did so at a Ptrans of 20 cm H₂O or less. Graphs of pressure-flow curves for three infants who showed different patterns of response are shown in Figure 4. Figure 4A is from an infant who had a past history of multiple severe wheezing episodes. It demonstrates an initial small increase in flows up to a Ptrans of 5 cm H₂O, followed by negative pressure dependence (i.e., a decrease with increasing pressure) in FEF_50% and FEF_75%. Figure 4B shows data from a healthy infant who demonstrated pressure independence in FEF_50% and FEF_75% between 7.5 and 25 cm H₂O, with negative pressure dependence in FEF_75% above 25 cm H₂O. Figure 4C is from an infant who, despite a history of wheezing episodes, had no evidence of pressure independence in any variable over the pressures employed. Curves similar to those in Figure 4B were the most common pattern seen. Regression lines of FEF on Ptrans were fitted for each individual infant at Ptrans above and below 20 cm H₂O. These are shown in Figure 5, and clearly demonstrate the likelihood of pressure independence at low lung volumes.

View larger version (20K): [in this window] [in a new window]   Figure 4.   Pressure-flow curves for three individuals who demonstrated different patterns of pressure dependence. (A) Early pressure independence and negative pressure dependence. (B) Pressure independence at Ptrans > 10 cm H2O in FEF50%, and FEF75%. (C ) No pressure independence. Flows are expressed as a percentage of baseline (i.e., as a percentage of flows generated by a Ptrans of 0). Error bars = SD

View larger version (20K): [in this window] [in a new window]   Figure 5.   Regression lines for effect of Ptrans on FEF25%, FEF50%, and FEF75%. Lines are fitted for the regression at Ptrans < 20 cm H2O and > 20 cm H2O. Clearly, pressure independence occurs more readily above 20 cm H2O and at low lung volumes.

In some infants, after the plateau in flows that occurred in response to increasing Ptrans, there was a secondary decrease in flow, demonstrating negative pressure dependence. This occurred in six of 16 infants (three healthy and three normal), at FEF_75%, in two of 16 infants at FEF_50%, and in none of 14 infants at FEF_25%. In only one infant was negative pressure dependence evident at Ptrans < 20 cm H₂O, and the curves for this are shown in Figure 4A. In all other cases negative pressure dependence was evident only at Ptrans in excess of 25 cm H₂O.

No clear differences between the healthy and wheezy groups of infants were seen in the patterns of pressure independence. Plots of group data for FEF_25%, FEF_50%, and FEF_75%, are shown in Figure 6A.

View larger version (19K): [in this window] [in a new window]   View larger version (19K): [in this window] [in a new window]   Figure 6.   (A) Pressure-flow curves with group data for FEF25%, FEF50%, and FEF75%. Flows are expressed as a percentage of baseline. (B) Plots of FEV0.5, FEV0.75, and FVC against Ptrans. Pressure independence is seen over the entire curve for FVC and at Ptrans > 20 cm H2O for FEV0.5 and FEV0.75. Error bars = SEM.

For the measurements made at Pi of 30 cm H₂O, a similar pattern emerged. In this instance there was evidence of pressure independence in one of nine infants at FEF_25%, in five of eight at FEF_50%, and in five of eight at FEF_75%.

The forced expiratory volume-time values of FEV_0.5 and FEV_0.75 also showed a very clear pattern of pressure independence at Ptrans > 15 to 20 cm H₂O. Group data from these variables are shown in Figure 6B, along with the curves of FVC plotted against Ptrans, which demonstrated pressure independent behavior over the range of Ptrans used. Once again, no clear differences between healthy and wheezy groups were seen. The results of the statistical modelling are presented in graphic form in Figure 7A and B for Pi values of 20 cm H₂O and 30 cm H₂O, respectively. The gradients estimated by the model for the effect of the predictor variable Ptrans on the outcome variables FEV_0.5, FEV_0.75, FVC, FEF_25%, FEF_50%, and FEF_75% are shown with the respective 95% confidence interval (CI) values. As previously stated, gradients were estimated for the early (Ptrans < 20 cm H₂O) and late (Ptrans > 20 cm H₂O) portions of the variable-Ptrans curve, and these are presented in Figure 7A and B, respectively. Group data for the wheezy and healthy groups are shown. Positive estimated gradients, with 95% CI values that do not include zero, indicate a statistically significant (p < 0.05) positive pressure dependence of the variable. Similarly, statistically significant negative estimated gradients indicate a negative pressure dependence. When the estimated gradient is associated with a 95% CI that contains zero, there is insufficient evidence to reject the hypothesis that the outcome variable is pressure independent. We can take this last result as demonstrating to some extent that the outcome variable is demonstrating pressure independence.

View larger version (19K): [in this window] [in a new window]   View larger version (20K): [in this window] [in a new window]   Figure 7.   Estimates of gradients with 95% CI for effect of Ptrans on the variables measured. (A) Inflation pressure = 20 cm H2O. (B) Inflation pressure = 30 cm H2O. (Upper panel  ) early gradients (Ptrans < 20 cm H2O). (Lower panel ) late gradients (Ptrans > 20 cm H2O). Gradients can be considered signifcantly (p < 0.05) different from zero where the error bars do not include the zero line.

A clear pattern emerges, consistent with the qualitative analysis, with significant positive early (Ptrans < 20 cm H₂O) gradients in all variables except FVC, for both levels of inflation pressure. Late (Ptrans > 20 cm H₂O) gradients with a Pi of 20 cm H₂O were not significantly different from zero in any variable except FEF_25% (positive), although the gradients for FEF_50% approached positive significance in both groups. The pattern of late (Ptrans > 20 cm H₂O) gradients in the group with a Pi of 30 cm H₂O was somewhat different. In this group there was a tendency for gradients to be significantly negative (FEV_0.75, FEF_75%, and FVC), particularly in the wheezy infants, suggesting that negative pressure dependence was more common over the late portion of the variable-Ptrans curve when a Pi of 30 cm H₂O was used.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We have presented strong evidence that there is a highly predictable range of driving pressures over which a majority of the variables measured with the RVFET will be pressure independent. This is important for two reasons. First, since minor variations in Pj, and consequently driving pressure, are invariable in any study, it is advantageous to be working on a flat (pressure independent) part of the variable Ptrans curve. Second, if driving pressures are routinely assessed, it will be possible to remove this important source of variation and allow improved comparisons between individual infants and between different populations or study groups. We have previously confirmed that it is possible, by simple and noninvasive means, to assess the contribution of the jacket to the overall driving pressure without using esophageal intubations (5, 9, 10).

It is important that the different pressures measured in this study be clearly understood. The overall driving pressure for expiration in the system is the transrespiratory pressure (Ptr). We measured this with an occlusion technique, at the raised volume set by the Pi (which equates to the static recoil pressure) of either 20 cm H₂O or 30 cm H₂O. This method is well validated, and correlates strongly with invasive measurements of dynamic changes in intrapleural pressure made with esophageal balloons (5, 9, 10). The Ptr measured in this way has two components, the first being the set Pi (20 cm H₂O or 30 cm H₂O) and the second being the Ptrans. The changes that occur in Ptr over the course of a raised volume forced expiration were studied previously (9, 10). There was a predictable decrease in Ptr over the tidal part of the curve, due to mechanical properties of the chest. However, pressures transmitted at lung volumes above end inspiration remained constant. Thus, it is not unreasonable to assume that, although the Ptr at any given FEF_% will be less than at end-raised inspiration, this Ptr will have a fixed relationship to the Ptr at end inspiration that will remain constant within a study. This assumption would be invalid if variable expiratory muscle activity were a major determinant of the changes that occur in Ptr over the course of expiration. However, this was shown not to be the case in a previous study (9) done with the rapid thoracic compression test (RTC) in paralyzed subjects.

The values obtained for the FVC of 21.5 ml · kg¹ at a Pi of 20 cm H₂O and 28.5 ml · kg¹ at a Pi of 30 cm H₂O are less than the value reported in the literature of 46.1 ml · kg ¹ (range 27.2 to 61.5 ml · kg¹) for newborn infants, achieved with the forced deflation (FD) method (8). This is probably due to differences in the pressures used to generate FVC (+40 cm H₂O to 40 cm H₂O with the FD method).

Because FVC is largely independent of Ptrans, except at very low pressures, it follows that the flows measured at fractions of FVC are effectively isovolumetric. For the low lung volume variable of FEF_75%, there was a clear portion of the pressure-flow curve in which pressure independence was demonstrated in most but not all of our subjects. This strongly suggests that flow limitation was occurring at low lung volumes. For FEF_50% a similar pattern emerged albeit in a smaller proportion of our subjects. However, at the higher lung volume flow of FEF_25%, we were unable to demonstrate evidence of pressure independence in any subject. Previous work has shown that flow limitation occurs more readily at low lung volumes (1). Consequently, it is possible that we were generating a Ptr insufficient to produce pressure independence at these volumes. If this was the case, it seems unlikely that it will be possible to achieve pressure independence over the whole of the flow-volume curve with the raised-volume forced expiration technique, since there is a limit to the Pj values that can be practically applied.

An alternative explanation for the apparent lack of pressure independence at high lung volumes is that it occurs but is obscured by measurement artefact. Two artefacts are more common at high lung volumes. The first is glottic closure, a phenomenon that is well known (8, 11) but about which there is little information in the literature. In our experience, it tends to occur early in expiration and therefore at high lung volumes. This may be due to reflex closure caused either by the high flows passing through the glottis, or in response to the rapidly inflating jacket. Glottic closure and noise on the signal caused by pharyngeal vibrations were more common with increasing Pj values. Chloral hydrate decreases upper airway tone (12, 13) and probably increases the likelihood of such phenomena. Although curves with notable glottic closure were excluded from our analysis, this phenomenon may have both obscured an effect and contributed to the increased variance seen at high Pj values.

The second artefact encountered at high lung volumes relates to the flow transients that occur early in expiration, and which are thought to be due to a combination of explosive decompression of gas in the upper airway and rapid inward acceleration of the upper airway wall (14, 15). These transients are accentuated by sudden decompression of the system (which includes the compliant upper airway) due to rapid opening of the main flow valve. These artefacts also tended to increase with increasing jacket pressure, and this increase could potentially obscure any pressure-(in)dependent behavior that was occurring.

We also present evidence of negative pressure dependence at low lung volumes with high Ptrans. This is further illustrated in Figure 8, with the flow-volume curves of a single infant at progressively increasing Ptrans values. After an initial increase, the FVC is constant. At higher Ptrans values there is clear evidence of a decrease in flow and scalloping of the curve at low lung volumes. This effect is clearly volume dependent, and is not seen at higher lung volumes. There are several possible explanations for this apparent negative pressure dependence. First, airway collapse is more likely to occur at low lung volumes, owing to decreasing support for the airway walls from the pulmonary parenchyma. This effect will be also accentuated by higher jacket compression pressures, leading to a pressure gradient across the airway that promotes collapse.

View larger version (16K): [in this window] [in a new window]   Figure 8.   Flow-volume curves of a single individual at progressively increasing Ptrans, demonstrating negative pressure dependence at high pressures.

Another factor contributing to an apparent negative pressure dependence could be the gas compression artefact (16, 17). This relates to the error generated by measuring expired volumes at the mouth with a pneumotachograph, instead of using a plethysmograph to measure expired volume. Gas compression and the error induced are pressure dependent, and could introduce an error on the order of 10% at high pressures. Errors from this effect are more likely to occur in individuals with airway obstruction, and this has been suggested as an advantage in distinguishing normal from abnormal subjects (18). However, it is also suggested (18) that these errors are affected by relatively small changes in effort (pressure in the present study), which would lend further weight to the argument for standardization of driving pressure.

The standard technique for generating max at FRC from partial flow-volume curves involves the use of progressively increasing pressure in the jacket, until no further increase in flow at FRC is seen (19). Applying this methodology to the RVFET is not without potential problems. Because we have shown that pressure independence occurs at different volumes at different pressures, it follows that the pressure required to achieve pressure independence at one volume may be insufficient to achieve pressure independence at a higher volume, but may in fact cause a negative pressure effect at a lower volume in the same curve. Obviously, if flow at only one fractional volume is reported, this difficulty does not arise. However, if it is intended to report flows at different fractional volumes, and/or integrated flows (FEV_t), there is potential for both under- and overestimation. Since we have shown that the Ptrans required to produce pressure-independent behavior of the parameters is between 15 and 25 cm H₂O, it should be possible to use a standard Ptrans in all infants to generate MEFV curves. We have studies more than 200 infants with this method, and have had no difficulty in producing Ptrans of 20 cm H₂O in any of them. This approach has certain advantages. First, if a standard driving pressure (Ptrans + Pi) is used in all infants, it will yield an estimate of the resistance to expiratory flow that will be comparable between infants even if pressure independence is not achieved over the whole expiration. This is not the case if different driving pressures are used in infants, nor with voluntary efforts in cooperative subjects since the driving pressure of the system under such circumstances is likely to vary between subjects. A second advantage is that, as has been previously suggested with the standard rapid thoracic compression (RTC) technique (20), use of a standardized pressure will allow improved comparison between curve shapes. Third, this approach would obviate the need to use high Pj values for some individuals. These advantages are all theoretical, however, and a head-to-head comparison of the two approaches (standardized versus individualized) would be required to confirm our belief.

Despite our collection of fewer data from infants at a Pi of 30 cm H₂O there seemed to be an increased likelihood of negative pressure dependence in these curves. There are two possible explanations for this phenomenon. First, the Ptr at the start of expiration will, when using a Pi of 30 cm H₂O, be 10 cm H₂O higher than the pressure when using a Pi of 20 cm H₂O for any given Ptrans. This would tend to increase the likelihood of negative pressure dependence. However, the higher Pi also leads to a higher starting lung volume and therefore a larger FVC. This means that FEF_25% measured at a Pi of 30 cm H₂O will be at a higher absolute lung volume than FEF_25% measured at a Pi of 20 cm H₂O, which would be expected to decrease the likelihood of achieving pressure independence. Alternatively, this phenomenon could be artefactual. Leaks around the mask are more common with increasing pressure in the mask, and occur most commonly just prior to expiration, when the pressure in the mask is at its highest. We have found that these leaks are more common with increasing Pj, and if this is the case it could explain the apparent decrease in flows with increasing Ptrans. Because we excluded any curves made when there was a leak noted either on the pressure trace or by the mask holder, this second explanation seems less likely.

The alinearity of transmission of Pj shown in Figure 3 is due to the characteristics of the jacket rather than to any physiologic property of the chest wall. We came to this conclusion by measuring Ptrans at the chest wall in a model, and demonstrating an identical pattern (data not shown). In summary, we have presented strong evidence of pressure independence over the latter half of expiration with the RVFET when transmitted pressures in excess of 20 cm H₂O are used. We have shown that it is possible to assess transmission of pressure from the jacket to the airway noninvasively with an occlusion technique, and that there is a range of such pressures that are likely to cause pressure independence over the latter half of expiration. We recommend that transmission of pressure be measured and reported in all studies, and suggest that if a standard Ptrans is used, it should be 20 cm H₂O.