Dead space and CO₂ elimination related to pattern of inspiratory gas delivery in ARDS patients

Ventilator-induced lung injury is an important problem in the acute respiratory distress syndrome (ARDS). It may be caused by barotrauma related to high airway, alveolar, and transpulmonary pressures or by shear forces at lung collapse and opening during tidal breaths. Among efforts to provide lung-protective ventilation in ARDS, a reduction of tidal volume (V_T) is a central issue [1‐5]. By using lower than traditional V_T, both of the mentioned damaging mechanisms may be mitigated. Recently, Bruhn et al. [6] showed by dynamic CT that cyclic collapse and opening is reduced by lower V_T, providing direct evidence that low V_T ventilation may be lung protective by reducing this phenomenon. However, the decrease in minute ventilation induced by low V_T can be difficult to offset by increasing respiratory rate and may induce hypercapnia. So, adequate CO₂ elimination under well-controlled airway pressure and tidal volume (V_T) is an important clinical issue, particularly in ARDS.

Accordingly, dead-space reduction plays a role in a rational lung-protection strategy. This suggests an optimal pattern of inspiratory gas delivery [7‐10]. Characteristics of this pattern are the time for gas insufflation (T_I), the postinspiratory pause time (T_P) after T_I, and the inspiratory flow wave pattern, denoted SHAPE. SHAPE may be constant: that is, square flow, increasing; or accelerating flow or decreasing that is decelerating flow. T_I, T_P, and SHAPE affect mean distribution time (MDT), that is, time available for distribution and diffusion of inspired tidal gas with resident alveolar gas [6‐8]. The definition of MDT is recapitulated in Figure 1. A longer T_P prolongs MDT and enhances CO₂ elimination [11]. It reduces dead space in healthy pigs and in pigs with acute lung injury [12]. The positive effects were more closely related to lnMDT than to MDT. In ARDS patients, a long T_P reduces dead space and leads to reduced PaCO₂, without a clinically significant increase in intrinsic positive end-expiratory pressure (PEEP) or negative hemodynamic effects [7, 8]. In one study, the T_P was varied, but also the T_I and SHAPE. The results indicated that positive effects were related not only to a high value of MDT but also to an abrupt cessation of end-inspiratory flow (EIF) that follows from shortening T_I or using an increasing flow pattern [12]. The definition of EIF is explained in Figure 2e.

No previous study investigated how various combinations of T_I, T_P, and SHAPE affect CO₂ elimination in ARDS patients, and no previous study distinguished between effects of MDT and EIF. The objective of the present study was to test in ARDS patients the hypothesis that the effects of the inspiratory gas-delivery pattern on CO₂ exchange can be comprehensively described from effects related to MDT, and also from effects of EIF, and to quantify the influence of these variables.

Eight mechanically ventilated subjects (Table 1) fulfilled criteria for ARDS [13]. Exclusion criteria were as follows: younger than 18 years, presence of a chest tube, contraindication to sedation or paralysis, intracranial disease, and a PaO₂/F_IO₂ < 75 mm Hg. Sedation and neuromuscular blockade were achieved by continuous infusion of midazolam and atracurium. The Ethics Committee of French Intensive Care Society approved the protocol, which was part of another study looking at different PEEP settings and F_IO₂ [14]. This is the background behind the values of PEEP and F_IO₂ in individual patients. Patients' next of kin were informed and gave consent to the study and its publication.

The patients were ventilated at volume control (ServoVentilator 900 C with a mainstream CO₂ Analyzer 930; Siemens-Elema, Solna, Sweden). Each patient had an arterial line. For ordinary breaths during basal ventilation, T_I was 33%, and T_P, 10%. Set PEEP was 5 cm H₂O. This low level of PEEP reflects the initial settings of the previous study [14]. The patients were ventilated with an effective V_T of about 6 ml/kg ideal body weight [1]. Measured volumes were corrected with respect to gas compression in ventilator tubing. Individual ventilation parameters are given in Table 2.

The ServoVentilator was controlled by a personal computer that emitted analog signals to the socket for external control of ventilator function [15, 16]. By controlling respiratory rate and minute ventilation, the computer instantly controlled inspiratory flow rate and durations of inspiratory flow (T_I) and postinspiratory pause (T_P), while maintaining constant tidal volume [12].

The computer was programmed to perform six recording sequences, each comprising 12 consecutive breaths. Breaths numbers 3, 6, 9, and 12 were modified. In six different sequences, single breaths were modified with respect to T_I (0.4 to 2.0 seconds), T_P (0.0 to 0.9 seconds), and inspiratory flow wave form (SHAPE). In total, breaths with 20 different flow patterns were studied (Figure 2). As seen, SHAPE was square (constant flow) or triangular, with increasing or decreasing flow rate. Average values from the four ordinary breaths representing basal ventilation immediately preceding the modified breaths were used as reference for the modified breaths in each recording sequence. V_T, PEEP, and expiratory time were constant for all breaths.

Signals representing airway flow and fraction of CO₂ at airway opening in percentage (FCO₂) were fed to the A/D converter of a personal computer and sampled at 100 Hz [12, 16]. Recorded data were transferred to an Excel workbook for analysis (Microsoft Corp., Redmond, WA, USA).

MDT was calculated from the average time that consecutive fractions of inspired gas remained in the respiratory zone, from their arrival in the zone until start of expiration (Figure 1) [7]. Further analysis was based on the single-breath test for CO₂, SBT-CO₂. SBT-CO₂ allows calculation of airway dead space, essentially from the maximal rising slope of the curve, as described in detail by Åström et al. [17]. When PaCO₂ is measured in steady state, alveolar dead space also can be determined [18]. Both dead-space fractions are affected by a modified inspiratory flow pattern, but only airway dead space can be studied at transient variations of ventilation. Therefore, in this study, the main studied variables were the volume of CO₂ eliminated during a single tidal breath (V_TCO₂) and particularly the change in V_TCO₂ (ΔV_TCO₂) in breaths modified with respect to inspiratory flow pattern (Figure 3). ΔV_TCO₂ was expressed in percentage of V_TCO₂ of ordinary breaths in the same recording sequence (ΔV_TCO₂%). A positive ΔV_TCO₂ or ΔV_TCO₂% indicates enhanced CO₂ elimination.

On average, 110 breaths were analyzed per patient, equal numbers of ordinary reference breaths and modified breaths with all combinations of T_I, T_P, and SHAPE (Figure 2). MDT varied between 0.12 seconds and 1.45 seconds, and EIF between 0 at decreasing flow and up to 1.7 L/sec, at increasing flow.

The computer-controlled ventilator allowed elaborate modification of inspiratory flow wave pattern on a breath-by-breath basis at constant tidal volume. Within 20 minutes of study time, about 20 types of modified breaths were studied. The ΔV_TCO₂% expresses how an alternative inspiratory pattern affects V_TCO₂ relative to ordinary breaths in the same recording sequence. Thereby, changes in CO₂ exchange due to variable physiological conditions between patients and in a single patient with time were minimized. With this technique and focusing on ΔV_TCO₂, we could for the first time comprehensively describe and distinguish effects of MDT and EIF in ARDS patients.

The most uncomplicated breath modification is when only T_P was modified, as this does not affect EIF. The findings confirm that the ΔV_TCO₂% is significantly affected by the length of the pause and tightly positively correlated to lnMDT [7, 12]. A comprehensive analysis of variation in T_I, T_P, and SHAPE has previously not been performed in patients. Such analysis showed that both MDT and EIF influence the V_TCO₂ and that the effects can be described with the simple equation:

At decreasing flow, more inhaled gas reaches alveoli early during inspiration. Thereby, MDT is prolonged. This will tend to enhance CO₂ exchange. However, EIF and MDT are negatively correlated to one another for breaths varied with respect to SHAPE. For example, for breaths with decreasing flow, MDT is long, but EIF is zero. This implies that EIF will balance and obscure effects of MDT; in agreement with that, no correlation between ΔV_TCO₂% and MDT was found among breaths with varying SHAPE.

Gas transfer over the boundary zone between fresh inhaled gas and resident alveolar gas and gas mixing within the alveolar zone are complex phenomena. Diffusion is a strictly time-dependent phenomenon and is believed to be the main phenomenon behind effects of MDT in ARDS. Diffusion drives gas transfer through the whole respiratory zone between capillary blood and conducting airways. In 1970, Knelson et al. [12] reported that a postinspiratory pause led to a more efficient alveolar gas exchange. They showed theoretically that this is due not only to improved distribution of ventilation but also to alveolar perfusion during the pause during which alveolar CO₂ tension approaches that of mixed venous blood. Fletcher et al. [19] further developed this concept. Aboab et al. [14] reasoned that a higher level of the alveolar plateau associated with a long pause might partly be due to continuing delivery of CO₂ by alveolar perfusion. In their study, a detailed analysis of breaths with similar MDT but variable distribution between T_I and T_P led to the suggestion that time for alveolar perfusion during inspiration is of low importance compared with time for distribution and diffusion within the alveolar zone, as expressed by MDT. Still, delivery of CO₂ to alveoli during T_I and T_P may to some extent contribute to positive effects of decreasing flow and a long T_P.

In a theoretic study, Jansson and Jonson [20] showed that a decreasing flow in combination with postinspiratory pause is favorable with respect to an even distribution of ventilation in the presence of uneven airway obstruction, a phenomenon associated with so-called pendelluft. They also postulated a small opposite effect in lungs with uneven compliance. In ARDS, effects on gas distribution to different lung regions are complex. Nevertheless, the influence of the inspiratory pattern is thought to relate more closely to diffusion within lung units than to distribution of ventilation between lung units.

Effects of EIF may reflect that high-frequency flow and pressure transients at airway opening are transmitted through the airways to the periphery of the lungs. This is the basis behind impulse oscillometry used for diagnostic purposes [21, 22]. In mechanical ventilation, it also is important for high-frequency oscillation [23]. Sudden interruption of flow at the end of inspiration, as expressed by EIF, implies that oscillations covering a broad spectrum of frequencies are transmitted through the airways down to the alveolar zone. Such oscillations serve to mix gas in the boundary zone between conducting airways and alveoli. This is the conceivable mechanism for the observed positive effect of a high EIF on CO₂ exchange.

The effect of the MDT on the ΔV_TCO₂% was similar among subjects (Figure 4, Table 3). In contrast, the effect of EIF varied importantly among subjects, as shown by large variation in coefficient b. The basis behind impulse oscillometry is that transmission of flow oscillations through the airways reflects differences in distributed resistance, compliance, and inertia along the airways. As lung mechanics is complexly perturbed in ARDS, the finding that EIF is of variable importance among patients could be expected.

In the present study, CO₂ exchange was studied only for single modified breaths at a time. It has been shown that a change in V_TCO₂ measured with the present technique will be followed by a corresponding change in PaCO₂ in the opposite direction [8, 10, 12, 24]. At constant respiratory rate, an observed change in V_TCO₂ after a change in tidal volume or inspiratory flow pattern will affect PaCO₂ in animals and in humans, in health and in disease. The layout of this study, based on studies of single breaths, has the strength of making it possible to study a large spectrum of inspiration and pause patterns within a short period, during which the physiological status of the patient remains essentially stable. Conversely, further studies are indicated in which particular patterns of inspiration are studied in steady state to evaluate effects on arterial blood gases, physiological dead space, and other parameters.

The results show that it is possible to enhance CO₂ elimination by about 12% to 15% just by modifying T_I and T_P at constant V_T in most ARDS patients and to about 20% in some patients (Table 4). At long T_P, further enhancement is possible by using increasing inspiratory flow. On modification of T_I and T_P leading to a higher I:E ratio, we could until now only partly understand clinical observations of immediate very important increments in CO₂ elimination observed on the 930 CO₂ Analyzer, followed by corresponding decrease in PaCO₂ after a switch to a more efficient pattern of inspiration. On the basis of the present results, an I:E ratio of 1:1 with a long T_P seems appropriate.

In most ARDS patients, it is more important to reduce V_T than to enhance CO₂ elimination. This would be the case in the present material in which hypercapnia was not a problem. Hypercapnia may have a lung-protective effect in itself. Such effects are not proven to improve outcome in ARDS [25]. However, hypercapnia combined with an efficient pattern of inspiration will allow particularly low tidal volumes, which may enhance lung protection. A combination of methods and approaches is needed for optimal reduction of dead space. Jonson et al. [26] showed that expiratory flushing of airways, later denoted tracheal gas injection [27], may be used to clear the airways from CO₂ down to trachea. To avoid potential problems of humidification of injected gas and of jet streams in the trachea, aspiration of dead space (ASPIDS) was developed and tested in animals and patients [28, 29]. With ASPIDS, dead-space gas is aspirated during late expiration through a catheter at the tip of the tracheal tube and simultaneously replaced by an equally large flow of fresh gas through the ordinary inspiratory pathway, avoiding all other influences on ventilation. In a porcine ARDS model, one can, with a combination of ASPIDS and the MDT concept, achieve normocapnia at very low V_T ventilation, as shown by Uttman et al. [10]. Despite the high metabolic rate in adolescent pigs, V_T was 4 ml/kg body weight. Notably, this very low V_T was achieved at respiratory rates of about 80 breaths/min. At high respiratory rates, MDT becomes shorter. Then, it is particularly important to choose a ventilation pattern that is optimal with respect to CO₂ exchange, which is a short T_I, a long T_P, combined with a short expiratory phase. This may augment intrinsic PEEP, an effect that should be balanced by a reduction of set PEEP. The study of Uttman et al. [10] illustrates that a reduced dead space paves the way for a higher respiratory rate. Theoretically, at dead space approaching zero, V_T could be reduced toward zero at very high respiratory rates.

Right-to-left intrapulmonary shunt contributes to alveolar dead space. As recently shown, this contribution increases at high metabolic rate, low cardiac output, low hemoglobin concentration, metabolic acidosis, and respiratory alkalosis [30]. Accordingly, conventional critical care measures aiming at homeostasis have the advantage of dead-space reduction. By addressing all meansfor dead-space reduction, in combination with much higher respiratory rates and much lower tidal volume than conventionally applied, adequate CO₂ elimination may be achieved in ARDS patients. Eventually one may reduce the use of extracorporeal gas exchange. The field for research remains wide open

CO₂ exchange at different inspiratory patterns can be described according to a simple equation based on MDT and EIF. Just by setting the ventilator to a pattern that enhances CO₂ exchange, one may reduce dead space and significantly increase CO₂ elimination or alternatively reduce V_T. This option merits use in clinical routine and, particularly, in further studies of optimal ventilation in ARDS.