Introduction

In acute respiratory distress syndrome (ARDS), addition of tracheal gas insufflation (TGI) to high-frequency oscillation (HFO) improves gas exchange [1, 2]. TGI may increase carinal pressure and promote lung recruitment [1, 2], improve the washout of the anatomical dead space [2, 3], and augment HFO-dependent, distal gas transport [1–3]. TGI flow of 6.0 L min−1 increases mean tracheal pressure (P tr), even if a cuff leak is used [1]. During standard HFO, P tr increases after a simple clockwise turn of the ventilator’s mean airway pressure (P aw) valve [1]. This may obviate any need for TGI.

We determined the effect of adding TGI to HFO on gas exchange independently from P tr. We compared standard HFO and HFO-TGI matched for a higher and a lower, clinically relevant P tr level [higher P tr level = 3 cmH2O above P tr of preceding conventional mechanical ventilation (CMV); lower P tr level = P tr of preceding CMV; see Electronic Supplementary Material (ESM)]. The rationale for additionally varying P tr was high-pressure-associated recruitment of dependent and well-perfused lung regions. This may improve ventilation–perfusion matching and promote TGI-enhanced distal gas transport [1]. A possible, TGI-related oxygenation improvement [1] could be greater at higher P tr versus lower P tr. Also, a higher P tr could increase physiological dead space, thus altering the effect of TGI on CO2 removal [1]. Consequently, potential differences in gas exchange between HFO and HFO-TGI could depend on P tr level.

Methods

Patients

Institutional approval and informed, written next-of-kin consent were obtained. Patients had early (onset within ≤72 h) acute lung injury or ARDS [4]: PaO2/FiO2 <250 mmHg with positive end-expiratory pressure (PEEP) ≥8 cmH2O for >12 h. Eligibility criteria, sedation, analgesia, neuromuscular blockade, and patient monitoring are detailed in the ESM. Patients received lung-protective CMV; PEEP and FiO2 were set in concordance with the ARDSnet protocol (ESM).

Study design

A randomized, crossover design was adopted. Figure 1 illustrates the study protocol. Additional details are provided in the ESM.

Fig. 1
figure 1

Schematic representation of the study protocol. CMV conventional mechanical ventilation, RM recruitment maneuver, HFO high-frequency oscillation, TGI tracheal gas insufflation, P aw mean airway pressure, P tr mean tracheal pressure. * Comprised hemodynamics, gas exchange (arterial and central-venous blood gases), and respiratory mechanics during CMV, and hemodynamics and gas exchange during HFO/HFO-TGI. † Comprised application of continuous positive airway pressure of 45 cmH2O for 40 s. ‡ For day 1, use of a P aw that was either “high” (i.e., titrated to a P tr that was 3 cmH2O higher relative to the P tr of the preceding CMV) or “low” (i.e., titrated to a P tr equal to the P tr of the preceding CMV) was randomized (www.randomizer.org); if day 1 P aw was “high,” day 2 P aw was low, and vice versa; this was done to control for the possible effect of disease evolution on our measurements. § During each day, the order of standard HFO and HFO-TGI was initially randomized and then reversed during the “repeat protocol”; this was done to control for any potential influence of the first HFO/HFO-TGI sessions on gas-exchange results obtained during the immediately subsequent HFO-TGI/HFO sessions

Full size image

Baseline CMV period

Tracheal tube correct positioning and patency were confirmed, and anesthesia and paralysis were induced. Within 15–20 min post enrollment, blood gases and respiratory compliance were determined. Subsequently, two 4.8-cm-long circuit adapters with side-arms were introduced in between the connector of the tracheal tube and the Y-piece of the breathing circuit (associated equipment dead-space increase = 12.5 mL). A rigid-wall TGI catheter [inner diameter (ID) = 1.0 mm, outer diameter = 2.0 mm] was passed through the proximal adapter. The tip of the TGI catheter reached the tip of the tracheal tube. A second catheter was passed through the distal adapter to monitor P tr. The tip of the pressure-measuring catheter was placed at 2.0 cm beyond the tip of the tracheal tube. The adapter/catheter system was maintained in place during the study intervention periods (Fig. 1). Following recruitment and 55 min of CMV, physiologic measurements were conducted (Fig. 1).

HFO strategies

The Sensormedics 3100B ventilator was connected to the proximal adapter. The TGI catheter was connected to a gas-mixing flowmeter for precise control of the composition of the TGI O2/air mixture. The TGI gas was not humidified, because the use of TGI was intermittent and brief [1]. High-frequency ventilator settings were: FiO2 = FiO2 of the preceding CMV, bias flow = 40 L/min, frequency (f) = 3.5 Hz, pressure amplitude (ΔP) = 90 cmH2O, and inspiratory-to-expiratory time (I:E) ratio = 1:2 (inspiratory time ~0.1 s). On HFO initiation, a recruitment maneuver was performed, a cuff leak of 4–5 cmH2O was placed, and P aw was titrated to the prespecified P tr level (Fig. 1). For HFO-TGI initiation, the recruitment maneuver was repeated, the cuff leak was placed, TGI (FiO2 = FiO2 of preceding CMV; flow = 50% of immediately preceding CMV minute ventilation) was added to HFO, and P aw was titrated to the prespecified P tr level (Fig. 1) [1]. During HFO/HFO-TGI, f and ΔP were to be maintained unchanged unless PaCO2 and pHa could not be kept within ±20 mmHg of the preceding CMV PaCO2 and >7.15, respectively. Within 55–60 min following HFO/HFO-TGI initiation, physiologic measurements were conducted (Fig. 1).

CMV periods following HFO/HFO-TGI

After 120 min of HFO/HFO-TGI, patients were returned to pre-HFO CMV. Within 55–60 min thereafter, physiologic measurements were repeated and the adapters/catheters system was removed (Fig. 1).

Mean esophageal pressure (P es)

In the last five patients, we measured P es as an indicator of lung volume (ESM).

Statistical analysis

According to an a priori power analysis, 22 patients were required for a = 0.048 and power = 0.86. Physiological data were compared using repeated-measures analysis of variance (ANOVA), followed by the Tukey test. Significance was set at P < 0.05. Data are reported as mean ± SEM.

Results

Twenty-two patients were enrolled within a 4-month period (Table 1; ESM). The effect of the catheters on the inspiratory resistance of the tracheal tube (mean increase: 41%) is detailed in the ESM.

Table 1 Patient characteristics and baseline ventilatory settings (volume-assist control mode with square-wave inspiratory flow)
Full size table

Ventilatory variables

Data are presented in Table 2. A lower (by 1–2 cmH2O) P aw was required during HFO-TGI versus standard HFO to achieve the prespecified P tr levels; the corresponding average peak tracheal pressure was also lower (by 2.3–2.7 cmH2O), whereas the average minimal (trough) pressure was 1.3–1.8 cmH2O higher, suggesting a TGI-induced PEEP effect [1, 2]; in five patients, P es was 1.6–4.7 cmH2O higher, indicating a TGI-induced increase in mean lung volume of ~68–139 mL (ESM). P tr was ~4 cmH2O higher during HFO/HFO-TGI-high P aw versus HFO/HFO-TGI-low P aw, because the reference P tr of the preceding CMV (Fig. 1) exhibited a variation of ~1 cmH2O.

Table 2 Ventilatory parameters and respiratory compliance during the study period
Full size table

Gas exchange

PaO2/FiO2 was higher during HFO-TGI-high P aw versus HFO-high P aw (281.6 ± 15.1 versus 199.0 ± 15.0 mmHg; mean increase: 42%; P < 0.001), and during HFO-TGI-low P aw versus HFO-low P aw (222.8 ± 14.6 versus 141.3 ± 8.7 mmHg; mean increase: 58%; P < 0.001) (Fig. 2). PaCO2 was lower during HFO-TGI-high P aw versus HFO-high P aw (45.3 ± 1.6 versus 53.7 ± 1.9 mmHg; mean decrease: 16%; P = 0.037), and exhibited a trend toward a lower value during HFO-TGI-low P aw versus HFO-low P aw (51.4 ± 1.7 versus 60.0 ± 1.9 mmHg; mean decrease: 14%; P = 0.09). At both P tr levels, pHa was higher during HFO-TGI versus HFO (P < 0.001) (Fig. 2). In five patients, the increase in PaO2/FiO2 and decrease in PaCO2 from HFO to HFO-TGI correlated with the estimated concomitant increase in mean lung volume (r 2 = 0.58–0.71, P ≤ 0.011; ESM, Fig. E5).

Fig. 2
figure 2

Results on gas exchange. FiO 2 inspired O2 fraction, CMV conventional mechanical ventilation, HFO high-frequency oscillation, TGI tracheal gas insufflation, P aw mean airway pressure, CMV pre-/post-high CMV preceding/following HFO-/HFO-TGI-high P aw, CMV pre-/post-low CMV preceding/following HFO-/HFO-TGI-low P aw. Filled squares diamonds, and circles represent mean value, and bars represent standard error of the mean. a Refers to the titration of P aw of HFO/HFO-TGI to either a higher (high P aw) or lower (low P aw) level of mean tracheal pressure (see also “Methods” and Fig. 1). *P < 0.05 versus preceding CMV. † P < 0.05 versus both HFO-high P aw and HFO-TGI-high P aw. § P < 0.05 versus HFO-high P aw. # P < 0.05 versus HFO-TGI-high P aw. ‖ P < 0.05 versus HFO-low P aw. P < 0.05 versus HFO-TGI-low P aw. **P = 0.09 versus HFO-low P aw

Full size image

Hemodynamics and CO2 elimination

Ventilatory technique did not affect hemodynamics; shunt fraction was lowest during HFO-TGI-high P aw (Table 3). During CMV and HFO/HFO-TGI, determinants of metabolic rate (e.g., temperature), medication (e.g., vasopressors/inotropes), oxygen consumption, respiratory quotient, cardiac index, and central-venous CO2 concentration remained stable (Table 3; ESM), indicating stable CO2 production and delivery rate to the lungs. Consequently, changes in PaCO2 reflected changes in CO2 elimination efficiency.

Table 3 Hemodynamics, central-venous oxygen saturation, and shunt fraction
Full size table

Discussion

HFO-TGI versus HFO resulted in higher PaO2/FiO2 (at both P tr levels) and lower PaCO2 (at the higher P tr level). HFO-TGI-induced, proportional blood-gas improvements were comparable between the employed P tr levels. These results imply enhanced lung recruitment and/or gas transport, and alveolar ventilation during HFO-TGI [1, 2, 5].

TGI impedes the opposite-directed expiratory flow, thus exerting a PEEP effect [3]; this explains the higher trough tracheal pressure, despite a lower P aw setting (Table 2). TGI-induced PEEP should increase expiratory lung volume and driving (i.e., alveolar) pressure, with consequent higher expiratory flow and less expiratory airway narrowing or closure [6, 7]. In five patients, the estimated TGI-induced increases in mean lung volume were explanatory of 58–71% of the TGI-induced improvements in gas exchange at both P tr levels. This suggests that lung recruitment is a major mechanism of TGI action. An increase in lung volume augments the axial wall tension of collapsed airways and facilitates their reopening [8], and decreases the diffusional resistance to gas exchange [1].

Preceding imaging studies [9, 10] showed that, at f ≤6 Hz, there is preferential distribution of ventilation in dependent and basal lung regions. Gas streams created by oscillatory flows tend to travel with minimal directional change, favoring lung base filling, since the branching angles of basal airways are less acute [11]. As steady flows behave similarly [12], the addition of TGI to HFO should further enhance lung base aeration, and ventilation–perfusion matching, since the vertical perfusion gradient is maintained during HFO [13]. Basal lung aeration is augmented as HFO tidal volume (V T) and corresponding pressure swings increase [9]. Our f and ΔP settings correspond to V T of ~200 mL [14]. The average, TGI-induced increase in V T was ~11 mL (~5%) (see ESM, Appendix II). TGI may have also acted as an “additional, high-velocity bias flow bypassing the resistance of the tracheal tube and augmenting the effect of a relatively large V T on CO2 elimination” [15].

The calculated mean molecule velocity of the TGI gas at the tip of the TGI catheter was 142.7 m/s, as opposed to an HFO mean inspiratory velocity of ~35.3 m/s at the tip of a tracheal tube with ID of 8.5 mm (ESM, Appendix II). During inspiration, the TGI jet stream may increase the difference between inspiratory and expiratory velocity profiles, and enhance convective gas exchange [16]. The high jet stream velocity implies short dead-space transit time and contribution to direct alveolar ventilation [17]. In the convection–diffusion zones, TGI flow may augment the asymmetry in inspiratory velocity profiles, promote the radial mixing and gas exchange with the expired alveolar gas, and enhance the airway bifurcation phenomenon [17]. Τhe relative importance of the mechanisms of TGI action during HFO warrants further study.

Limitations

Limitations of routine, long-term TGI include absence of commercially available equipment, and possible tracheal mucosal damage, retention and inspissation of secretions, hemodynamic compromise, pneumothorax, and gas embolism [1, 3]. Humidification of TGI gas and cuff leak during HFO-TGI are essential [18]. We excluded patients with 7.5–8.0-mm ID orotracheal tubes, because the two catheters would raise inspiratory tube resistance to >20 cmH2O/L/s (ESM). For long-term TGI, we employ just one catheter, causing an increase in inspiratory tube resistance of <20%, and administer humidified TGI gas [18; ESM].

Clinical implications

When added to CMV, TGI improves CO2 clearance, increases total PEEP, and enables reduction of V T and driving pressures [19, 20]. This may facilitate lung protection in ARDS patients with poor tolerance to hypercapnia due to brain edema, or concurrent metabolic acidosis [19, 20]. In such clinical scenarios, HFO-TGI may prove superior to standard HFO.

Targeting gas-exchange improvements through increases in P aw or V T may accentuate ventilator-associated lung injury. However, HFO-TGI-low P aw versus HFO-high P aw resulted in similar gas exchange (Fig. 2) at lower ventilator-applied lung distending pressure(s), i.e., P aw and P tr (Table 2).

Conclusion

At equal P tr level, HFO-TGI results in superior gas exchange compared with HFO.