Introduction

A considerable number of patients suffering from severe brain damage require mechanical ventilation during the acute posttrauma phase. Although their morbidity and mortality are caused principally by the primary disease, medical complications are frequent, with respiratory dysfunction being the most common nonneurological organ system failure [1]. In spite of this no study has assessed respiratory mechanics on the 1st day of mechanical ventilation in brain-damaged patients. Moreover, although these patients need prolonged mechanical ventilation because of coma and neuroprotection, no study has followed these patients to assess whether mechanical ventilation-related parameters affect respiratory mechanics.

It has been shown that brain-damaged patients undergo a profound inflammatory response characterized by the release of several cytokines [2] and neuropeptides [3]. The biological effects of these substances include leukocyte infiltration and expression of adhesion molecules [4], bronchoconstriction, mucosal edema, and increased vascular permeability [5]. There is also evidence that some of these substances influence the expression of pulmonary surfactant proteins [6]. Furthermore, traumatic brain injury in animals increases lung vulnerability to subsequent injurious insults [7] and also causes ultrastructural damage in type II pneumocytes [8] which may result in impaired function of pulmonary surfactant.

In healthy adults reduced functional residual capacity, peripheral airway closure, atelectasis in the dependent lung zones and impairment of gas exchange with an increased alveolar-arterial oxygen tension difference (DA-aPO2) are common during general anesthesia in supine position [9, 10]. Moreover, the presence of airway closure and atelectasis implies inhomogeneous expansion of the lungs with development of heterogeneous mechanical stresses and peripheral airways injury; with positive end-expiratory pressure (PEEP) no such injury is observed [11, 12, 13].

This study assessed respiratory mechanics on the 1st and 5th days of mechanical ventilation in a group of brain-damaged patients on zero PEEP (ZEEP) or PEEP of 8 cmH2O, the hypothesis being that on ZEEP heterogeneous lung mechanics (atelectasis, airway closure, tidal expiratory flow limitation) worsens respiratory dysfunction while PEEP reduces or prevents it altogether. The results provide the first mechanics data of brain-damaged patients during early (1st day) mechanical ventilation on ZEEP. Part of this study has been previously reported in the form of an abstract [14].

Materials and methods

Subjects

The study protocol was reviewed and approved by our institutional ethics committee. Informed consent was obtained from next to kin. Twenty-four consecutive patients admitted to our ICU were recruited according to the following criteria: (a) diagnosis of severe brain damage defined as Glasgow Coma Scale lower than 8 on admission after resuscitation [15]; (b) no acute lung injury (ALI), as estimated by a PaO2/FIO2 ratio higher than 300 and normal chest radiography [16]; and (c) intracranial pressure (ICP) higher than 8 mmHg. An ICP level above 8 mmHg was chosen to prevent the transmission of PEEP through the cerebral veins according to the Starling resistor model [17]. ALI patients were excluded since their management mandates application of PEEP. Brain damage was diagnosed by clinical history, neurological examination, and brain computed tomography. Exclusion criteria were hemodynamic instability and previous history of cardiopulmonary disease. Of the 24 patients who met the inclusion criteria 21 completed the study: in one patient cerebral perfusion pressure fell below 70 mmHg, while two patients became septic (all from the ZEEP group). Patients' characteristics are summarized in Table 1; the two groups did not differ significantly regarding these.

Table 1 Anthropometric characteristics, gas exchange data, and cause of brain-damaged patients (ICP intracranial pressure, ICP ZEEP ICP on zero PEEP, ICP PEEP ICP after the application of PEEP)
Full size table

Clinical management and study protocol

Mechanical ventilation was initiated in all patients at the emergency room of our hospital. Upon transfer to the ICU patients were sedated using continuous infusion of midazolam or propofol and fentanyl, paralyzed with cisatracurium if needed, and nursed in supine position with an approx. 30° head tilt. All patients were ventilated by volume control modality (Servo 900 C, or 900E, Siemens-Elema, Solna, Sweden). Tidal volume and respiratory frequency were targeted to maintain a PaCO2 equal to 30–35 cmH2O. Patients were randomly assigned to receive 8 cmH2O of PEEP or ZEEP. Respiratory system mechanics and blood gases were measured on 1st and 5th days of ICU stay. During this period PEEP was kept constant whereas FIO2 and minute ventilation were modified by the attending physician if needed. Arterial pressure (using standard pressure transducers), ICP (CaminoV-420, Camino medical products, San Diego, Calif., USA), electrocardiography, heart rate, and pulse oximetry were continuously monitored (Life Scope 14, Nihon Kohden, Tokyo, Japan). Patients with signs of sepsis as defined by the American College of Chest Physicians/Society of Critical Care Medicine Consensus Conference criteria [18] were excluded from the study. A physician not involved in the study was always present to provide for patient care.

As standard care, treatment for increased ICP was initiated at an upper threshold of 20 mmHg, consisting of osmotic diuretics, hyperventilation, and hypertensive therapy provided that complete volume resuscitation had been assured, aiming to a constant cerebral perfusion pressure higher than 60 mmHg [19]. To assure brain protection patients whose cerebral perfusion pressure decreased below 70 mmHg were withdrawn from the study. If patients were judged as stable, sedation was discontinued for short periods for assessment of brain function.

Instrumentation and measurements

Flow was measured using a Hans-Rudolph pneumotachograph (model 3700A) with a linearity range of  ± 2.6 l/s attached at the proximal end of the endotracheal tube and connected to a differential pressure transducer (DP 55,  ± 3 cmH2O; Raytech Instruments, Vancouver, B.C., Canada). Pressure at airway opening was measured from a port inserted between the endotracheal tube and the pneumotachograph, connected to a pressure transducer (DP 55,  ± 100 cmH2O) via a rigid polyethylene tube (1.7 mm ID). Pressure transducers were calibrated before each study. Lung volume changes were obtained by numerical integration of the flow signal. Data analysis was performed using Direc (version 2.18; Raytech Instruments) and Anadat (version 5.2; RHT-InfoDat; Montreal, Que., Canada).

Respiratory mechanics was assessed by the standard constant-flow airway occlusion technique [20] with the ventilator settings reported in Table 2 plus a 3-s end-expiratory occlusion followed by a 5-s end-inspiratory occlusion. Ventilator settings did not differ significantly between the two patient groups. In addition to daily care, suctioning was performed 15 min prior to each respiratory mechanics measurement. No patient received bronchodilation therapy. During measurements the humidifier was omitted from the ventilatory circuit. Static elastance (Est,rs), additional resistance (Δ Rrs), maximum resistance (Rmax,rs) and minimal resistance (Rmin,rs) of the respiratory system were computed as previously described [21]. Rmax,rs and Rmin,rs were corrected for the endotracheal tube resistance [22].

Table 2 Ventilatory settings of brain-damaged patients (VT tidal volume, VTml/kg tidal volume per kilogram of ideal body weight, V̇E minute volume, V̇I inspiratory flow, TI,TTOT inspiratory and total breathing cycle time, F respiratory rate)
Full size table

Statistical analysis

Data of respiratory mechanics are presented as mean  ± SD; comparisons between experimental conditions (PEEP vs. ZEEP) were performed with the t test while those between days 1 and 5 were performed with the paired t test. Regression analysis used the least squares method. Analyses were performed using the software package Statistica. A p value less than 0.05 was considered statistically significant.

Results

On day 1 FIO2 was 0.46 ± 0.10 in both groups and remained essentially constant during the 5-day study period (day 5: 0.45 ± 0.10; Table 3). On day 1 there were no significant differences in gas exchange or respiratory system mechanics between the two groups (Table 3). On day 5 there was a significant PaO2/FIO2 decrease in both groups; in the ZEEP group significant increases in PaCO2, Est,rs and iso-CO2 Rmin,rs (see below) occurred and an almost significant increase in DA-aPO2 (p = 0.059). By day 5 one patient under ZEEP exhibited ALI.

In the ZEEP group there was a significant increase in PaCO2, which is known to affect airway caliber [23]. Both on ZEEP and PEEP we found a significant correlation between Rmin,rs changes (Δ Rmin,rs) and the corresponding PaCO2 changes (Δ PaCO2) between days 1 and 5 (Fig. 1). When the effects of the PaCO2 changes on Rmin,rs [23] were taken into account using the regression models (Fig. 1), on day 5 there was a significant increase in Δ Rmin,rs (intercept at Δ PaCO2 = 0) on ZEEP (Δ Rmin,rs = 1.5 ± 0.6; p = 0.04) but not on PEEP (1.0 ± 0.9; p = 0.29). Based on this analysis, under iso-PaCO2 conditions (i.e., at the PaCO2 of day 1) the values of Rmin,rs on day 5 were 7.0 ± 1.9 cmH2O/l per second in the ZEEP group and 6.5 ± 3.2 in the PEEP group. Values for iso-CO2 Rmin,rs were significantly higher on day 5 than on day 1 in the ZEEP group but not in the PEEP group.

Fig. 1
figure 1

The relationship of the change in minimal resistance of the respiratory system (Δ Rmin,rs) between days 5 and 1 to corresponding change in PaCO2 (Δ PaCO2) in ten brain-damaged patients ventilated on ZEEP (left) and 11 patients ventilated on PEEP (right). All changes refer to absolute values

Full size image
Table 3 Respiratory mechanics and gas exchange data of the two groups of patients (FIO 2 fraction of inspired oxygen, Est,rs static elastance of respiratory system, Rmax,rs maximum resistance of respiratory system, Rmin,rs minimal resistance of respiratory system, iso-CO 2 Rmin,rs minimal resistance of respiratory system corrected for corresponding changes in PaCO2 from day 1 to day 5, Δ Rrs additional resistance of respiratory system)
Full size table

A significant correlation was found between Δ Rmin,rs/Δ Est,rs, an index of change in the time constant of the respiratory system, and Δ PaO2/FIO2 in the ZEEP group (r = −0.8, p = 0.006) but not in the PEEP group (Fig. 2). In the ZEEP group a significant correlation was also found between Δ Rmin,rs and Δ DA-aPO2 between day 5 and day 1 (r = 0.66, p = 0.039). ICP values were similar in ZEEP and PEEP groups (day 1, 18 ± 5 vs. 16 ± 2 mmHg, p = 0.58; day 5, 17 ± 3 vs. 18 ± 5 mmHg, p = 0.71). Similar values were obtained in both groups throughout study period.

Fig. 2
figure 2

The relationship of the change in PaO2/FIO2 between days 5 and 1 to corresponding change in minimal resistance over change in static elastance of the respiratory system (Δ Rmin,rs/Δ Est,rs) in ten brain-damaged patients ventilated on ZEEP. All changes refer to absolute values

Full size image

Discussion

The main findings of this study are (a) in brain-damaged patients without ALI on day 1 of mechanical ventilation on ZEEP the static elastance and the minimal resistance were above normal limits; (b) a further increase in static elastance and minimal resistance of the respiratory system was found on day 5 of mechanical ventilation on ZEEP but not with PEEP of 8 cmH2O; (c) one of the ten patients on ZEEP developed ALI; (d) on day 5 there was a significant decrease in PaO2/FIO2 in both patient groups; in the ZEEP group there was an increase in DA-aPO2 which approached significance (p = 0.059).

Day 1

Although on day 1 our patients did not have ALI, their PaO2/FIO2 was below normal limits [24]. This is probably associated with airway closure and atelectasis that are common with sedation/anesthesia [9, 10]. Their PaCO2 was low since the brain-damaged patients were therapeutically hyperventilated to produce cerebral vasoconstriction and reduce both brain water and blood flow [19]. The ZEEP group exhibited intrinsic PEEP (PEEPi) up to 3.4 cmH2O in spite of the low tidal volume and long expiratory time used during mechanical ventilation (Table 2). In normal anesthetized paralyzed subjects ventilated at ZEEP with similar tidal volume (7 ml/kg) and expiratory time (2.6–4.4 s) PEEPi is absent [25]. In the PEEP group PEEPi was predictably absent [26].

Est,rs was higher than in normal, anesthetized, paralyzed subjects, as reported by D'Angelo et al. (18.9 ± 3.8 vs. 14.5 ± 2.1 cmH2O/l; p = 0.001) [25]. This discrepancy cannot be attributed to size differences because height and body mass index were similar, nor to the small, although significant, age difference between the two groups (31 ± 7 vs. 39 ± 13 years; p< 0.002) [27]. Pulmonary edema could partially explain the increase in elastance. Sympathetic hyperactivity with intracranial hypertension or free radicals produced as a result of central nervous system injury may elicit neurogenic pulmonary edema [28]. However, there was no evidence of cardiovascular instability or radiographic findings consistent with pulmonary edema in any patient. Hence atelectasis which is commonly associated with anesthesia and supine position in normal subjects [10], was probably the main contributor to the increase in Est,rs in our patients. Atelectasis and peripheral airway closure [9, 10] were probably enhanced by impaired production or function of pulmonary surfactant due to brain damage [6, 8], contributing to the increased Est,rs.

Rmin,rs was higher than in normal, anesthetized, paralyzed subjects (5.6 ± 2.2 vs. 2.3 ± 0.5 cmH2O/l per second; p< 0.001) [25]. This discrepancy should be due in part to different PaCO2 in normal and brain-damaged patients (37 vs. 31 mmHg). In normal anesthetized, paralyzed subjects a decreased PaCO2 is associated with a significant increase in Rmin,rs [23]. Additional factors may include bronchoconstriction and mucosal edema caused by neuropeptides, such as substance P, which are released in patients with brain damage [3, 5].

Δ Rrs values in our patients were higher (although not significantly) than those computed at the same inflation time (1.05 s) from data obtained by D'Angelo et al. [25] in normal subjects (4.1 ± 2.2 vs. 3.2 ± 0.7 cmH2O/l per second; p = 0.1). This is not due to differences in PaCO2 since Δ Rrs is independent of this parameter [23]. The increased Δ Rrs in the brain-damaged patients may reflect increased time constant inequality [21] leading to increased DA-aPO2, but it may also reflect changes in viscoelastic tissue behavior.

Respiratory mechanics has been assessed in few previous studies although not on the 1st day of mechanical ventilation. Tantucci et al. [29] studied five patients with brain damage at unspecified time after onset of mechanical ventilation on ZEEP. Their Rmin,rs values and PEEPi were similar to those of the present study. In contrast, Est,rs tended to be lower than in the present ten patients ventilated on ZEEP (Est,rs: 13.3 ± 7.1 vs. 18.9 ± 3.8 cmH2O/l, p< 0.06). This probably reflects to large extent the high tidal volume used in that study (15 ml/kg). Caricato et al. [30] provided respiratory mechanics data of brain-damaged patients ventilated on ZEEP but without specifying either the day of mechanical ventilation or ALI presence/absence. Mascia et al. [31] assessed Est,rs in brain-damaged patients with ALI and found values higher than in the present study (24.5 vs. 18.9 cmH2O/l, respectively).

Day 5

In the PEEP group there were no significant changes in respiratory mechanics between days 1 and 5, and no patient developed ALI. In contrast, one patient on ZEEP exhibited ALI by day 5. Furthermore, allowing for the changes in PaCO2 on day 5, there was a 25% increase in Rmin,rs (p < 0.006) in the ZEEP group. However, assuming that this increase was due entirely to peripheral airways (see below), and that peripheral airway resistance represents 20% of Rmin,rs [32], this resistance may have trebled. Moreover, in the ZEEP but not the PEEP group Est,rs significantly increased in relation to day 1.

The changes in mechanics in the ZEEP group cannot be attributed to pulmonary overdistention because the values of end-inspiratory plateau pressure were lower than in the PEEP group, and well below 30 cmH2O [33]. Absorption atelectasis is an unlikely cause since in both groups low FIO2 was used during the 5-day study period. Accordingly, deterioration in respiratory mechanics from day 1 to day 5 in the ZEEP group probably reflects “low volume” injury due to airway closure or heterogeneous constriction [11, 12, 13, 34, 35, 36, 37].

General anesthesia and supine position promote atelectasis in the dependent lung zones and peripheral airway closure even in normal lungs [9, 10]. In our subjects aged 39 ± 13 years, the closing volume should be well above the end-expiratory volume [9]. This implies opening and closing of peripheral airways during tidal breathing and development of shear stresses that can damage peripheral airways. In normal open-chest rabbits mechanical ventilation at low lung volume with physiological tidal volume leads to histological and functional alterations characterized by damage of terminal bronchioles, rupture of alveolar-bronchiolar attachments, and increased airway resistance [11, 12]. The latter was found in our patients on ZEEP. In the above studies on open-chest rabbits D'Angelo et al. [11, 12] found no change in Est while in our patients on ZEEP Est,rs was increased on day 5 of mechanical ventilation. However, in a subsequent study on closed-chest animals both airway resistance and Est were found to be significantly increased after 3 h of mechanical ventilation at low lung volume [13].

In the presence of airway closure there is heterogeneous lung filling and emptying, which probably also contribute to lung injury [26, 37, 38]. In patients with brain damage abnormal surfactant production resulting from pneumocytes II damage [8] or release of inflammatory mediators [6] could enhance peripheral airway closure and atelectasis.

Although significant, the increase in Est,rs and Rmin,rs found in our patients from day 1 to day 5 was relatively small probably because they were young and nonobese. Old age and obesity are known to promote airway closure [9]. Thus in elderly obese subjects as well as in patients with diseases such as chronic obstructive pulmonary disease [39] and chronic heart failure [40], which promote airway closure, protective PEEP may be mandatory. In our brain-damaged patients a modest level of PEEP (8 cmH2O) provided protection against lung injury, probably by restoring lung volume and reducing lung heterogeneity.

It has been shown that PEEP application does not induce a significant reduction in cerebral perfusion pressure if ICP values are higher than the PEEP applied [41]. In contrast, it has been shown that in brain-damaged patients with ALI application of PEEP lower than ICP increases the latter only by inducing pulmonary hyperinflation resulting to a significant PaCO2 increase [31]. Given that in our patients PaCO2 was always low, a CO2-mediated increase in ICP was unlikely. Despite similar PaCO2 values, in the PEEP group minute ventilation was slightly higher than in the ZEEP group. Since this difference was not statistically significant, increased dead space ventilation in the former group does not seem likely. Therefore application of moderate PEEP levels in selected brain-damaged patients is safe provided that euvolemia is guaranteed and continuous monitoring of ICP is available. Whether PEEP has any benefit on outcome in this setting remains unproven.

Conclusions

The present study shows that on day 1 of mechanical ventilation patients with brain damage exhibit abnormal respiratory mechanics. After 5 days of mechanical ventilation on ZEEP there are significant increases in Est,rs and Rmin,rs which perhaps reflect “low volume” injury, and which could be prevented by administration of moderate levels of PEEP.