Expiratory flow limitation detected by forced oscillation and negative expiratory pressure

The presence of airflow obstruction is a defining feature of several lung diseases and its persistence over time and despite bronchodilator treatment is typical in chronic obstructive pulmonary disease (COPD) 1. When airflow obstruction develops, lung emptying is delayed during forced manoeuvres and there is a reduction in the ratio of forced expiratory volume in one second to forced vital capacity, a change which also occurs with aging but to a lesser degree 2. As airflow obstruction worsens, expiratory flow limitation (EFL) appears at much lower flows for a given lung volume and it becomes present at rest or at least develops early during exercise 3. This is important as changes in dynamically regulated lung volume that are characteristic of exercise in COPD are likely to be directly related to the presence of EFL 4. However, initial methods of detecting EFL proved either invasive, involving balloon catheterisation, difficult to standardise because of variations in previous lung volume history, or involved relatively complex and problematic plethysmographic techniques.

The development of the negative expiratory pressure (NEP) technique provided a relatively simple way of identifying flow limitation by comparing the expiratory flow–volume profile of a control breath to that of a breath when additional negative pressure of ∼3.5–5 cmH₂O was applied 5. Any increase in flow beyond the control data demonstrates that some expiratory flow reserve is present. This method is independent of volume and time history, it is noninvasive and relatively simple to apply in a variety of clinical settings (e.g. in the intensive care unit and during exercise) and it has been used in COPD patients 6–9 and in other respiratory and systemic disorders 10–14. There is a possibility of upper airway artefacts in some patients and standardised methods of interpreting NEP data have not yet been published, although individual investigators are familiar with these practical issues 8, 11.

An alternative approach has recently been described in which within-breath changes in respiratory system reactance (ΔX̄_rs) measured by forced oscillation technique (FOT) were used to identify flow-limited breaths 15, 16. There was clear separation between flow-limited and non-flow-limited breaths when this method was compared with data from balloon catheters.

These approaches are based on different physical principles. NEP detects the condition in which all possible pathways between airway opening and the alveoli are choked. When this occurs, the total expiratory flow is independent of the expiratory pressure, a condition of “global” EFL. In contrast, FOT assesses the amount of the lung that is choked during expiration only. This measures “regional” flow limitation, and a threshold value indicates when the regional flow limitation reaches the condition of global flow limitation. Therefore, when global EFL is reached, the two techniques should produce the same response.

In the present study these methods have been used to determine whether a given breath was flow limited. It was hypothesised that NEP and FOT methods would produce similar results when directly compared, despite the different approaches adopted to identify EFL. To test this idea, NEP and FOT methods were compared within the same breath, along with the impact each has on classifying an individual patient as being flow limited. In addition, it was examined whether bronchodilator drugs, which conventionally improve lung emptying, modify EFL detected by either method.

METHODS

RESULTS

The characteristics and lung function data of the COPD patients are shown in table 1⇓. Figure 3⇓ shows a representative time course of volume, P_ao, respiratory resistance, X_rs and ΔX̄_rs for a few breaths before and after a NEP manoeuvre obtained with the experimental set-up used in the present study. This patient was classified as flow limited by ΔX̄_rs. Using the raw data the loop shown in figure 4a⇓ was obtained. After the filtering procedure, the same manoeuvre produced the plot in figure 4b⇓. This last graph was used for the NEP scoring. Altogether, 180 breaths were studied in this way. Of these, 105 were recorded in the 21 patients under baseline conditions and 75 were recorded from the 15 patients who repeated the test after bronchodilator. On average, each observer discarded 24.8% of the flow–volume loops because they did not meet the acceptance criteria. Altogether, 11.2% of discarded breaths met criterion 1, 18.6% met criterion 2, 14.2% met criterion 3 and 55.2% met criterion 4.

DISCUSSION

The detection of EFL during tidal breathing is a potentially important measurement which has been substantially simplified by the development of the NEP and, more recently, the FOT measurements described in the present study. Both methods define EFL independently of the previous volume history of the test, are noninvasive and easily repeated. Despite these similarities, each method exploits different physical principles to identify EFL, which might affect their ability to classify individual breaths or individual patients as being flow limited. The present data, in which the methods are compared using the same breath, are reassuring but highlight several factors that can influence the categorisation of individual breaths and patients as being flow limited or not.

The protocol of the present study compared the two methods using the same breath to measure ΔX̄_rs and as a reference breath for the subsequent NEP application. Thus, any possible effect of FOT on breathing pattern, which has not previously established, would affect both methods to a similar degree. The FOT method uses empirically derived criteria for identifying flow limitation, which are applied automatically. As the decision to classify a given breath as being flow limited using the NEP method could be influenced by the observer, three independent observers who were unaware of the ΔX̄_rs data were used to score the NEP traces. A set of rules to determine EFL by the NEP method was developed, something which has not been explicitly stated in previous reports. This is the first occasion when a comparison of blinded interobserver agreement has been reported for the NEP method.

To compare quantitative results between FOT and NEP the degree to which the expiratory flow volume loops of the control and NEP breaths overlapped was scored. However, applying NEP produces an additional expiratory flow from the upper airways (fig. 2⇑) making it impossible to determine whether the initial period of the breath is flow limited or not. The use of a lower NEP might reduce this artefact but this phenomenon cannot be avoided entirely. As the duration of the artefact depends on the amplitude of the applied pressure and the extrathoracic airways compliance, this introduces a variability in the scores that is unrelated to the degree of flow limitation. For this reason it was decided to arbitrarily assign a score of 100% (breath flow limited) to the breaths in which flow limitation is clearly present as soon as the artefact disappears.

A NEP pressure of ∼7 cmH₂O was applied, which is somewhat greater than that usually used. This might have increased the number of breaths discarded because of upper airway collapse artefacts, as suggested by other workers 11, 22. Although this increased the number of loops discarded by the observers it did not affect the reliability of NEP or change the relationship between NEP and FOT in the remaining tests; in fact, only two evaluations out of 36 were not possible because all the breaths were discarded by all the observers. Finally, to permit appropriate comparison with the control breath, a special filtering procedure to electronically subtract the FOT fluctuations superimposed on the flow–volume loops was developed.

Although both the methods detect the presence of EFL, the physical principles they use to do so are different and this may contribute to some of the discrepancies in classification that were observed. During EFL, the impedance measured by FOT is a measure of the mechanical properties of airways downstream from the choke points. This is because a change in pressure cannot be transmitted upstream through the choke points and only the downstream airways are oscillated 23. As airway wall compliance is one order of magnitude greater than lung compliance, the reduction of X_rs during expiration reflects the number of choke points that occurred and their distribution within the bronchial tree. Therefore ΔX̄_rs reflects the overall distribution of flow limitation within the lung, and the threshold indicates the value above which all the pathways between airway opening and the alveoli are choked. Conversely, NEP technique detects only the latter condition, i.e. when all pathways are choked, in which further increase in alveolar pressure cannot lead to increase in expiratory flow.

This difference could have an important impact on the clinical meaning of the measurements provided by the two methods. For example, ΔX̄_rs starts increasing with the progressive development of choke points, associated with a decrease in expiratory flow reserve. In these conditions, some airways are not yet flow limited and would increase their flow by increasing alveolar pressure, thus making the phenomenon go undetected by NEP.

Moreover, as FOT provides a “quantification” of number and location of choke points, very different values for ΔX̄_rs were found within the fully flow-limited patients (2.89–15.20 cmH₂O·s·L⁻¹). Even if the clinical meaning of the value of ΔX̄_rs has still to be investigated, bronchodilation significantly affected ΔX̄_rs, suggesting that the degree of EFL has been modified, although the changes were too small to modify the classification of the patient to not flow limited for most of the patients flow limited at baseline. This information cannot be provided by any method to detect EFL based on the comparison of expiratory flows.

All these differences might explain the higher sensitivity that ΔX̄_rs showed in assessing the effects of bronchodilators, which could change the number of airways in which the choke points occur or cause the choke points to move distally within the airway without necessarily changing the NEP score.

Another important difference between FOT and NEP is that FOT does not depend on the evaluation of an observer but provides a number obtained by an automatic algorithm. This might be useful especially for the classification of borderline patients which, in the present study, showed the largest disagreement between observers when using NEP.

Given all these differences between the methods, the comparison of the results can obviously be performed by considering their ability to detect full EFL, which was the main goal of the present study. In these conditions there was good agreement between the methods in identifying flow-limited and non-flow-limited breaths when the observers all agreed about the scoring. The small number of false-positive breaths reported in this comparison was largely the result of using a single threshold value of ΔX̄_rs to classify flow limitation rather than the range of data noted in the original empirical study 15. In the majority of cases where there was disagreement between NEP and ΔX̄_rs in the whole dataset there was also disagreement between the observers about the NEP scores, suggesting that these breaths were difficult to score using the NEP method.

These methods were also used to determine whether individual patients had EFL by averaging the NEP scores from all the observers for a given patient. Since the number of averaged NEP scores was variable and there were always five ΔX̄_rs scores available, the comparison between the average of NEP scores and ΔX̄_rs values might be statistically biased. However, it reflects the way in which the two techniques are performed in clinical practice and, therefore, the current authors believe that it provides a sensible approach for the comparison of the two methods.

In the present analysis, two sources of variability were observed in the test results: breath-to-breath variation in the degree of EFL within a patient and interoperator differences in the scoring of NEP loops. Both methods identified between-breath changes in the degree of flow limitation in the same patient, a process likely to reflect spontaneous variation in dynamically regulated end-expiratory lung volume. Data presented in figure 6⇓ support this view, where ΔX̄_rs over several minutes in three different patients illustrates both breath-to-breath variability and the way in which an individual close to the threshold value for flow limitation (fig. 6c⇓) can change between the two states. FOT appears best suited to detect these changes as NEP requires at least 45 s of stable breathing between measurements, which makes following flow limitation in a dynamic manner more difficult. Differences in NEP scoring between operators meant that it was necessary to derive an empirical threshold for agreement between NEP scores and ΔX̄_rs. When the average NEP score was >30.8% there was agreement between NEP and ΔX̄_rs in determining EFL and, similarly, patients were reliably classified as not flow limited when the average score was below 24.6%. These empirically derived thresholds might differ if different observers were used. However, the improvement seen in the agreement between methods when the NEP data were averaged suggests that much of the variance lies in how the NEP are interpreted.

Bronchodilator drugs would be expected to modify the degree of EFL, and this has been proposed as an explanation of the reduced end-expiratory lung volume during exercise that usually 24, 25, but not always 26, follows bronchodilator treatment in COPD. However, studies to date have been disappointing, with no change in NEP score after the administration of high-dose bronchodilators to patients with severe COPD at rest 6. The present data confirmed these findings, three patients changing from flow limited to not flow limited using the ΔX̄_rs criteria, two of whom also showed a change in NEP scores. However, one patient increased the degree of EFL by both the methods. Overall, there was a significant reduction in ΔX̄_rs after bronchodilator, the clinical significance of which is still to be determined.

Both methods have good measurement properties and, thanks to the recent technological advances in digital and power electronics, similar complexity and production costs.

The subjective nature of the negative expiratory pressure response there is good interobserver agreement, especially when the breath is clearly flow limited or not. In contrast, forced oscillation technique is relatively “objective” and the values derived from balloon catheter data distinguishing flow-limitated from non-flow-limited breaths are in good agreement with the classification of flow limitation using the negative expiratory pressure method. The ability to measure multiple breaths over longer periods and to do so automatically means that the within-breath change in reactance method is well suited to continuously monitoring expiratory flow limitation, which may be desirable in the intensive care unit or when an intervention such as noninvasive ventilation is planned 16. Data obtained with both methods suggest that the degree of flow limitation varies from breath to breath in patients with stable chronic obstructive pulmonary disease. Thus, individuals can move through a transition state where flow limitation may or may not be detected in an individual breath. This emphasises the need for testing several breaths in the evaluation of a patient′s flow-limitation status. Further studies of the factors that determine this variability are now possible using the forced oscillation technique.