INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The electrical activity recorded from a muscle constitutes the spatial and temporal summation of action potentials from the recruited motor units, provided that neuromuscular transmission and muscle fiber membrane excitability are intact. The electrical activity of the crural diaphragm (EAdi) should therefore be a valid reflection of phrenic nerve activity (1), and, hence, of neural drive to the diaphragm. However, when measured with an esophageal catheter, EAdi is recorded from a small sample of the total number of crural diaphragmatic motor units that are activated. This signal may therefore be criticized as not being representative of total diaphragm activity.

Transdiaphragmatic pressure (Pdi) is an index used often to represent the activity of the entire diaphragm. Whereas EAdi represents the neural drive to the diaphragm, Pdi is a consequence of that neural drive. Unlike the electrical activity, Pdi is dependent on the neuromechanical coupling, that is, the transformation of electrical activity into pressure generation. Some of the relevant factors that influence neuromechanical coupling include both dynamic and static changes in lung volume.

We have previously demonstrated in healthy subjects a relationship between crural EAdi and global diaphragm activation (Pdi) during static (2) and dynamic (3) contractions under well-controlled experimental conditions and when changes in neuromechanical coupling were taken into account. Despite these findings, esophageal recordings of EAdi are not routinely used in the clinical setting to evaluate neural drive. For example, measurements of EAdi could be useful for the purpose of monitoring or adjusting respiratory support in mechanically ventilated patients.

The aim of the present study, therefore, was to compare crural and global diaphragm activity by looking at the changes in electrical activity of the crural diaphragm (EAdi) and the changes in Pdi during conditions of varying respiratory drive. The study was carried out in ventilator-dependent patients and, in order to promote changes in respiratory drive, patients were evaluated while breathing on different levels of pressure support ventilation (PS).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients

Thirteen intubated patients with ARF of varying etiology were studied. Their mean age, height, and weight (mean ± SD) were 70.92 ± 8.3 yr, 159.08 ± 10.33 cm, 64.54 ± 20.85 kg, respectively. (See Table E1 in the online data supplement for individual patient data and diagnoses.) The protocol was approved by the ethics committee of Hôpital Maisonneuve-Rosemont, Montreal, Quebec, Canada. Informed consent was obtained. Patients were minimally sedated and were breathing spontaneously while receiving PS. Patients were mechanically ventilated (Siemens 900C, Siemens, Solna, Sweden; or Puritan Bennett 7200; Mallinckrodt, Carlsbad, CA) with the a mean PS level of 14.46 ± 5.58 cm H₂O, and PEEP of 5.92 ± 2.4 cm H₂O. (See Table E1 in the online data supplement for individual patient data.)

Instrumentation

Flow was measured with a no. 2 pneumotachograph (Hewlett Packard; Palo Alto, CA) and a pressure transducer (Ohmega Engineering Inc., Stamford, CT; ± 3 cm H₂O). Volume was obtained by digitally integrating flow. Airway opening pressure (Pao) was measured from a side port of the endotracheal tube via a separate pressure transducer (Sensym Inc., Milpitas, CA; ± 350 cm H₂O). EAdi was recorded using a multiple array esophageal electrode, as previously described (3, 4). EAdi signals were amplified, filtered, acquired, and digitized as previously described (4). Esophageal pressure (Pes) and gastric pressure (Pga) were measured via balloons mounted on the same catheter and connected to two pressure transducers (Sensym Inc.).

Experimental Protocol

The catheter was positioned so that the most centered electrode pairs were positioned at the level of the crural diaphragm (5). Proper placement of the Pes balloon was confirmed using the occlusion test (6). Data were acquired while patients breathed on: (1) the prescribed PS setting for 5 min, and (2) randomly varying PS levels, altered every 5 min, in steps of multiples of 5 or 10 (or multiples of 7 in one patient). Externally applied PEEP was not adjusted.

Analysis

EAdi signals were processed with algorithms as previously described (4). The root mean square (RMS) of EAdi was calculated.

An averaged inspiration for the last 2 min of each period was calculated. Inspirations in which Pes was contaminated by esophageal spasms or other artifacts were excluded. The beginning and the end of the diaphragm activation period were visually determined as the point where EAdi started and ceased, respectively, and were selected as reference points for each individual inspiration. From the averaged breath, tidal volume (VT) and both mean and peak inspiratory values of both Pdi and RMS were calculated. Respiratory rate was calculated from flow. We also calculated the VT at peak of EAdi (VT_peakEAdi) (Figure 1) from the averaged breaths.

View larger version (16K): [in this window] [in a new window]   Figure 1.   Schematic representation demonstrating calculation of volume to peak of EAdi (VTpeakEAdi). Top panel shows EAdi tracings from averaged inspirations in Patient 3 at PS levels of 0 (thin solid lines) and 15 (thick solid lines) cm H2O. Middle panel shows the corresponding inspired volumes, and the bottom panel shows the corresponding transpulmonary pressure tracings. Arrows indicate the volume at which the EAdi peaked (dashed lines, PS = 15; dotted lines, PS = 0). PS = pressure support. EAdi = diaphragm electrical activity. a.u. = arbitrary units.

Inspiratory pulmonary resistance was determined using the elastic subtraction technique (9). Dynamic lung elastance was calculated as the quotient of the change in transpulmonary pressure (obtained by subtracting esophageal from airway opening pressure, at points of zero flow) and of inspired tidal volume. Changes in dynamic intrinsic PEEP were calculated from the changes in Pes from the onset of inspiratory Pes deflection to the point of zero flow and corrected for expiratory muscle activity (10).

Statistical Analysis

Pearson's product correlation was used to evaluate the relationship between EAdi and Pdi at multiple PS levels, within the individual subject. Student's paired t tests were used to compare the parameters at lowest and highest PS levels. Student's paired t tests were also used to evaluate whether or not RMS and Pdi changed with the addition of 10 cm H₂O PS were significant; p < 0.05 was considered significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Data in text and tables are presented as means and standard deviation (mean ± SD). All patients tolerated the experimental protocol and insertion of the catheter well. The mean percentage of rejected breaths in the calculation of the averaged breath was 18% (range, 0 to 54%).

End-expiratory Pdi values, as well as the Pdi swing for a given change in EAdi were unaltered with changing PS (p = 0.38 and p = 0.97, respectively). The breathing pattern changes indicated an increase in VT from 430 ml ± 180 to 527 ml ± 180 (p < 0.001) (mean difference, 97 ml ± 83; p < 0.001) and a reduction in respiratory rate from 23.9 ± 7 to 21.3 ± 7 breaths/ min (p = 0.015), when PS increased on average from 2 ± 4 to 20 ± 7 cm H₂O. The lowest pressure support level in 10 of the 13 patients was zero cm H₂O, i.e., CPAP at the set PEEP level. There was no significant change in minute ventilation (9.4 ± 2.4 to 10.2 ± 2.3 L/min). Mean inspiratory flow changed from 0.53 ± 0.18 L/s from low PS to 0.62 ± 0.18 L/s at high PS (p = 0.023) when calculated from the entire flow Ti period. There was also no consistent change in PEEPidyn (p = 0.489) with increasing PS (mean change was 0.17 cm H₂O ± 0.84 from the lowest to the highest PS levels, respectively) for the group.

Figure 1 shows in one patient (Patient 3) the lack of increase in VT_peakEAdi at lowest PS (dashed tracings PS = 0 cm H₂O) and highest PS (solid tracings = 15 cm H₂O), and the corresponding tracings of transpulmonary pressure. For the group, VT_peakEAdi was unaltered from the lowest to highest PS (p = 0.976), (276 ± 147 ml to 277 ± 162 ml). Transpulmonary pressure at peak EAdi increased from lowest to highest PS (18.41 ± 7.6 to 21.5 ± 5.6 cm H₂O), p = 0.168, with no changes in dynamic lung elastance (p = 0.261), and a slight increase in pulmonary resistance (p = 0.04). (See Table E2 in the online data supplement for individual patient data of VT_peakEAdi, transpulmonary pressure, elastance, and resistance at different PS levels.) The values for VT_peakEAdi, normalized for body weight, for all subjects at all PS levels, as well as the group mean data, are shown in Figure 2. Mean inspiratory flow calculated to the peak of EAdi was not different from low (0.49 ± 0.22) to high PS (0.64 ± 0.24), (p = 0.06).

View larger version (14K): [in this window] [in a new window]   Figure 2.   Inspired volume at peak EAdi, normalized to body weight, in all patients at all PS levels. Thin patterned lines indicate individual patient data (n = 13). Thick solid lines and symbols represent mean group data at lowest support (LS) and highest support (HS).

The RMS and Pdi tracings for the averaged inspirations obtained at different levels of PS in Patient 8 are shown in Figure 3. A progressive decrease in both EAdi and Pdi was observed with the addition of PS. For the group, the mean correlation coefficient between the mean RMS and mean Pdi for different levels of PS was r = 0.84 ± 0.12, and for the peak RMS and peak Pdi, the mean correlation coefficient was r = 0.90 ± 0.09, as presented in Table 1 for each individual subject.

View larger version (22K): [in this window] [in a new window]   Figure 3.   Example of averaged inspirations in one patient mechanically ventilated on four different pressure support levels (see legend ). Left panel shows diaphragm electrical activity (y axis), and right panel shows transdiaphragmatic pressure (y axis) plotted versus time (x axis). Pdi = transdiaphragmatic pressure. a.u. = arbitrary units.

In those patients in whom a 10 cm H₂O step change in PS was evaluated (n = 12), EAdi and Pdi group mean changes were compared. On average there was a 33% ( ± 21%) reduction in mean RMS values (p < 0.001) when PS was increased by 10 cm H₂O. This was the same as the decrease in mean Pdi (p < 0.001) observed (34% ± 36). There was no statistical difference between the change in the mean RMS and the mean change in Pdi values when PS was increased by 10 cm H₂O (Table 2). The same trends were observed for the peak RMS and peak Pdi values (reduction by 37 ± 23% and 35 ± 23%, NS).

In Figure 4 an averaged EAdi time profile for the entire group, at the lowest (dashed line) and the highest (solid line) PS levels, are shown. (Note that only three points of the diaphragm activation pattern are included, and this does not represent the true EAdi time-profile.) The timing parameters are shown in more detail in Table 3. For the group, the end of inspiratory flow coincided with the end of neural activity at high PS, as shown in Figure 4 (see arrows and dotted lines), and preceded the end of neural activity at low PS by 72 ms. EAdi at high PS levels peaked at 53.5 ± 23% of the value observed at low PS levels.

View larger version (26K): [in this window] [in a new window]   Figure 4.   EAdi profile for the group, for low (dashed lines, closed circles) and high (closed lines, open circles) PS levels. Three points are connected (beginning, peak, and end of diaphragm activation period) and do not represent the actual shape of the EAdi-time profile. Note that diaphragm electrical activity peaks lower and sooner at high PS levels. The beginning of ventilator assist and the end of inspiratory flow are indicated by arrows. PS = pressure support. EAdi = diaphragm electrical activity. a.u. = arbitrary units.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The present study has shown that esophageal recordings of crural EAdi can be used to determine changes in global diaphragm activation (i.e., respiratory drive) in a group of mechanically ventilated patients with acute respiratory failure. This was obtained when no evidence for changes in neuromechanical coupling were found (with varying levels of PS). It must be made clear that Pdi, unlike EAdi, is dependent on the neuromechanical coupling (2), and would not have been valid for comparison to EAdi if neuromechanical coupling had changed with varying PS.

Neuromechanical Coupling of the Diaphragm

In the present study, we use the term neuromechanical coupling of the diaphragm to describe the transformation of electrical activity (EAdi) into pressure generation (Pdi). Evidence that neuromechanical coupling was maintained at low and high PS was that Pdi at the onset of EMG was unaltered with changing PS, and the swing in Pdi for a given change in EAdi was also unaltered with changing PS. Considering that neuromechanical coupling of the diaphragm is affected by changes in both end-expiratory lung volume (EELV) and inspiratory volume, we evaluated these parameters during changing PS as well. Both of these alter the diaphragm radius of curvature, diaphragm length, and zone of apposition (11), thereby affecting the pressure generation of the diaphragm, for a given level of diaphragm activation (2). In the present study, no evidence for changes in EELV was observed with changes in PS, similar to the work of previous investigators (12). This was based on the fact that dynamic PEEPi did not change with changing PS. EELV could change while PEEPidyn remains constant if compliance changed in the opposite direction to compensate for this; however, we did not observe any significant changes in compliance with changing PS. In the present study, breathing frequency was reduced and VT was higher with increasing levels of PS, providing further support that EELV did not change.

During dynamic breathing, increasing volume causes a progressive neuromechanical uncoupling throughout the breath. Although the amount of uncoupling in relation to the magnitude of dynamic volume change remains to be evaluated, it has been shown during submaximal, static diaphragmatic contractions that increasing lung volume produces neuromechanical uncoupling (2). It is important to be aware that the relative changes in Pdi (i.e., the percentage change) with changes in muscle length are the same at submaximal or maximal levels of diaphragm activation (2, 14). Therefore, when comparing one inspiration with another, the degree of uncoupling will remain relatively similar, as long as the tidal volume is the same. However, when lung volume changes, the degree of neuromechanical uncoupling will change. In a study performed in patients with COPD mechanically ventilated with BIPAP, Viale and colleagues (15) demonstrated similar values of diaphragm electrical activity for spontaneous and assisted breaths, while on the other hand, Pdi was reduced and VT increased during the assisted breaths. Their data suggest an uncoupling between diaphragm activation and pressure generation simply caused by the delivery of mechanical assistance.

However, with increasing PS, we did not find any change in inspiratory volume during the period when diaphragm activation increased (VT_peakEAdi) from low to high PS levels (Figures 1 and 2 and Table E2 in the online data supplement). Therefore, in our study, similar intrabreath uncoupling between EAdi and Pdi likely occurred with changing PS during the period of increasing diaphragm activation, suggesting that the relationship between peak EAdi and peak Pdi was equally affected with changes in PS.

In contrast to the period of the breath when diaphragm activation increases, the issue of intrabreath uncoupling becomes more complex during the period when diaphragm activation decreases (and ventilator assist continues) since the volume-induced neuromechanical uncoupling continues to increase while EAdi and Pdi are decreasing. At both high and low PS levels, we found that the end of inspiratory flow occurred close to the time when neural activity returned to baseline, indicating that inspiratory volume reached its peak when the diaphragm was already deactivated. Consequently, when one would have expected the maximum impact of volume on neuromechanical coupling (i.e., at peak inspiratory volume) there was actually no more EAdi and Pdi. Considering the small difference between VT and VT_peakEAdi when comparing low and high PS, its uncoupling effect on the declining EAdi and Pdi is likely small. Although small, this uncoupling may explain the lower correlation coefficients observed between the mean EAdi and mean Pdi values (calculated for the entire period of diaphragm activation), compared with the peak EAdi and peak Pdi values.

Another factor that can impair neuromechanical coupling of the diaphragm is the force-velocity relationship of skeletal muscle (16). It has previously been shown in healthy subjects that the Pdi generating capacity for a given EAdi is maintained up to inspiratory flows of 1.4 L/s (3). In the current study, mean inspiratory flow rates were approximately less than half of this value, precluding the possibility of any flow-rate-related uncoupling between EAdi and Pdi with varying PS.

Does Crural EAdi Reflect Global Diaphragm Activity in Acute Respiratory Failure?

In the present study, we have shown that within an individual patient, RMS and Pdi changed in the same direction with changing PS, and crural diaphragm activation correlated with Pdi at different levels of PS. In general, these results suggest that sampling of a limited number of activated motor units in the crural diaphragm with an esophageal electrode can estimate changes in activation of the whole diaphragm in mechanically ventilated patients. This is consistent with our previous work in healthy subjects where we showed that crural diaphragm activation was related to global diaphragm activation, when changes in neuromechanical coupling were taken into consideration (2). We also showed that EAdi was unaffected by changes in lung volume, whereas Pdi was.

In a patient with high ventilatory demands (e.g., a patient in acute respiratory failure with high drive), we believe that the respiratory muscles are activated close to their maximum, and that there are few "degrees of freedom" with which the crural and costal portions can vary their activation patterns. In patients with stable COPD performing maximal voluntary diaphragm activation maneuvers (e.g., Pdi sniff) costal and crural diaphragm electrical activities correlated well (4). To our knowledge, no studies have yet been carried out in humans to examine the electrical activities of the costal and crural portions of the diaphragm during acute respiratory failure. In an animal preparation, similar increases in the electrical activities of the phrenic nerve, costal diaphragm, and intercostal muscles were reported during acute respiratory failure (17). The work of Lourenco and colleagues (1) also demonstrated a close relationship between phrenic nerve activity and either costal or crural diaphragm electrical activities during inspiratory loading in dogs.

Critique of the Methods

With increasing PS crural diaphragm activation was reduced with a concomitant reduction in Pdi. These findings are in agreement with previous work demonstrating reductions in various indirect measurements of respiratory "effort" with increasing PS (18). However, the purpose of the present study was not to evaluate the absolute changes in drive for a given change in PS, but rather to compare EAdi and Pdi under conditions of varying respiratory drive.

The mechanism of PS-induced changes in respiratory drive is unclear. Regardless, the group data showed that there was a similar reduction in global and crural diaphragm activation with the addition of 10 cm H₂O in PS; however, the individual responses to the addition of PS varied (range was 5 to 62% reduction for the mean RMS). It is possible that this variable response to the addition of PS was due to the patients being minimally sedated, the etiology of the ARF, the patients' individual respiratory mechanics, as well as the type of ventilator, the settings being used, and the order of sequential step changes. Despite the interindividual differences (whatever the cause), EAdi and Pdi correlated very well for a given subject (see Table 1), and the mean relative change in EAdi and Pdi for a step change of 10 cm H₂O was similar. Again, these results were found during conditions of no evidence for changes in neuromechanical coupling.

The present work used esophageal recordings of EAdi as a reflection of crural diaphragm activation. The crural diaphragm electrical signal strength is dependent on accurate methodology for acquisition (i.e., electrode configuration), analysis (5, 7) and electrode positioning (5). The signal strength is also influenced by factors affecting signal quality such as cardiac activity, electrode motion artifacts, and noise, technical problems that have been overcome in recent works (4, 8).

The use of esophageal recordings to evaluate diaphragm activation in acute respiratory failure is most often criticized because of claims that the signal strength is affected by changes in lung volume (15, 23, 24). We have demonstrated in healthy subjects that the RMS of the diaphragm electrical activity is not artifactually influenced by changes in chest wall configuration and/or lung volume during voluntary contractions (2). These results finally demonstrated that any increase in diaphragm electrical activity observed with an increase in lung volume is not an artifact in the signal, but rather a true increase in neural drive to the diaphragm that is required to compensate for the shortened length. The method has been validated in healthy subjects during both static (2) and dynamic (3) maneuvers, as well as in patients with diaphragm weakness (caused by COPD or polio), at rest (4) and during exercise (25).

Indications about Patient-ventilator Interaction

The results of the present study provide interesting information about patient-ventilator interaction and cycling off of mechanical ventilation during PS, when cycling off criteria are based on the flow signal. During unassisted breathing in healthy subjects, the period of increasing activation (up to the peak activity) is associated with inspiratory flow, and the period of decreasing activation is associated with the beginning of expiratory flow. Therefore, in our study, poor patient ventilator interaction was evident as almost half the tidal volume was still being delivered by the ventilator during the period of decreasing diaphragm activation, an observation that was further exaggerated at higher PS levels. Therefore, unless other respiratory muscles are producing inspiratory flow during the diaphragm deactivation period (which is unlikely), it seems that poor patient-ventilator interaction during conventional cycling-off of PS can be described by a continued delivery of volume during the relaxation period of the diaphragm.

In conclusion, the results of the present study have demonstrated that, in mechanically ventilated patients with acute respiratory failure, crural diaphragm electrical activity measured with an esophageal electrode represents global diaphragm activation. These findings suggest that esophageal recordings of diaphragm electrical activity can be useful for monitoring respiratory drive and for adjusting the magnitude of ventilatory assist during or between breaths.