Introduction

Acute respiratory distress syndrome (ARDS) is defined by radiographic diagnosis of diffuse bilateral alveolar infiltrates, the degree of hypoxemia, lung function, and histopathology. It is the final generalized inflammatory response of the lung to catastrophic events of various pulmonary and nonpulmonary origins and occurs in all age groups. The diagnostic criteria have been established by an American-European Consensus Conference [1, 2, 3]. ARDS in children is still associated with a high mortality. Mortality is correlated with the severity of the underlying disease [4, 5]; however, it is still generally accepted that the degree of hypoxemia predicts outcome [5, 6, 7, 8]. Mortality in children with a PaO2/FIO2 ratio lower than 100 in central Europe is still greater than 50% [5]. Another outcome criteria is the aggressiveness of ventilatory support, reflected in the peak inspiratory pressure (PIP), mean airway pressure, and ventilation index [9, 10]. The enforcement of standardized ventilation protocols and the lower incidence of both multiple trauma and sepsis in central Europe have caused a continuous decrease in severe ARDS incidence in children [11, 12, 13].

As the mortality is still high in children with profound hypoxemia and severe underlying conditions such as immunosuppression [4, 5, 8, 13], many other therapies in addition to baro- and volutrauma preventing ventilation strategies have been reported. Nevertheless, randomized controlled studies evaluating new therapeutic strategies for the treatment of severe hypoxemic ARDS as nitric oxide, high-frequency oscillatory ventilation (HFOV), extracorporeal membrane oxygenation (ECMO), and surfactant treatment are lacking in the pediatric age group [14, 15, 16, 17, 18]. Exogenous surfactant improves oxygenation in neonates not only with respiratory distress syndrome but also in other conditions with secondary surfactant deficit such as meconium aspiration syndrome and congenital pneumonia, comparable with ARDS [19]. Oxygenation improved after surfactant administration in several case reports on adult ARDS patients [11, 20] and in a controlled study in moderate pediatric ARDS [10]. A retrospective and prospective survey of pediatric ARDS in all major German pediatric intensive care facilities demonstrated that the mortality and chronic illness after ARDS in patients with PaO2/FIO2 ratios above 150 is very rare, and that most patients with a PaO2/FIO2 ratio around 200 are not even ventilated [5]. For this reason the German-Austrian working group on ARDS performed a controlled, randomized study in severe, hypoxemic ARDS in children, associated with a very high mortality, and no evidence-based treatment options available.

Patients and methods

A single-center pilot study was carried out in 19 pediatric patients with ARDS and PaO2/FIO2 ratio lower than 100, aimed at dose finding and calculating the number of patients required. This pilot study treated patients with ARDS defined by consensus conference criteria and aged between the 44th postconceptional week and 14 years of age. They were included if their PaO2/FIO2 ratio was below 100 for at least 2 h. Ventilation in the pilot study was in accordance with the ventilation algorithm established by the German Working Group on Pediatric ARDS (Fig. 1). A bovine surfactant (Alveofact, Boehringer, Ingelheim, Germany) was administered in 50 mg/kg increments (intratracheal bolus under continuous ventilation and PEEP over maximally 5 min) as long as the PaO2/FIO2 ratio did not increase by 20% or decrease by 10%. The PaO2/FIO2 ratio was determined at 48 h The increase/decrease from baseline was evaluated using the Mann-Whitney U test. No other exclusion criteria were applicable [17, 19]. This pilot study as the following mulitcenter study were approved by the ethics review board of the principal investigators' institution (Medical University of Lübeck).

Fig 1.
figure 1

Algorithm for ventilatory management in the study centers before and after randomization

Full size image

The subsequent multicenter study was an open, randomized, parallel group comparison performed at 19 German pediatric intensive care units between May 1997 and November 1999. These units were referral centers for children with severe respiratory failure. All children were randomized if they fulfilled the following criteria: ARDS Consensus Conference criteria, lung injury score [21] of at least 2, ventilation between 12 and 120 h, age between 44th postconceptional week and 14 years, admission for at least 4 h, no echocardiographically detectable left heart failure, and a PaO2/FIO2 ratio lower than 100. Written informed consent was obtained from the parents or legal guardians. Patients were excluded if they were under other investigational or experimental therapies: nitric oxide, high frequency ventilation with current disease, liquid ventilation, prostaglandins, steroids for therapy of ARDS, ECMO; chronic lung disease such as bronchopulmonary dysplasia or cystic fibrosis, participation in other clinical studies except treatment protocols and studies for oncological diseases. In addition, patients with severe hypoxemia, i.e., PaO2 lower than 50 after 4 h of treatment in the referral center were excluded.

Thirty-eight patients were recruited over the 18-month period (mean age 3.9 years), but three randomized patients were not included in the study, as their PaO2/FIO2 ratio improved to 100 or higher before the start of treatment. All patients in the participating centers with ARDS in the specified age group who were not included as they did not fulfill all entry criteria were recorded and evaluated as "intended to treat patients" if their PaO2/FIO2 ratio was below 100. All patients in the participating centers were ventilated according to a ventilation algorithm (Fig. 1) [4], blood pressure was kept above the 50th percentile for age, fluid intake between 90% and 100% maintenance [22], and hemoglobin levels above 12 g/dl. As the PaO2/FIO2 ratio was less than 100, patients were randomized either to receive 100 mg/kg bovine surfactant (Alveofact, Boehringer-Ingelheim, Germany), administered under continuous PEEP and ventilation as a bolus over not longer than 5 min to the distal tip of the endotracheal tube (n=20), or be continuously treated based on the ventilation algorithm and concomitant therapy outlined above (n=15). In the surfactant group an additional dose of 100-mg/kg surfactant during the 48-h observation period was allowed if the PaO2 decreased by 20% from the maximum level reached. Diagnoses, age, weight and other demographic characteristics are depicted in Table 2, including the ratio of immunosuppressed patients (after bone marrow transplant or under chemotherapy), PRISM III score, and pneumonia; there were no significant differences between the two groups in regard to these factors.

Table 1. Reasons for not randomizing patients with ARDS and PaO2/FIO2 ratio less than 100 ("intended to treat"). The survival rates were not compared statistically because of the small number of patients in each section of the intended to treat patients; the overall survival rate was 36%
Full size table
Table 2. Patient characteristics in surfactant and nonsurfactant groups
Full size table

PaO2, FIO2, paCO2, pH, peak inspiratory pressure, PEEP, ventilatory rate, tidal volume, blood pressure, heart rate, hemoglobin, all medications administered, fluid balance, and the derived variables (PaO2/FIO2, Hallman oxygenation index, ventilatory index, mean airway pressure) were recorded 2, 4, 12, 24, and 48 h after randomization. The patients were followed to day 30 or to discharge/transfer from the referral center to document outcome data: survival, rescue therapy, days on ventilator, days in the ICU, Pediatric Risk of Mortality (PRISM) III score [23], lung injury (Murray score) [21], and days in hospital.

Random allocation was performed centrally by telephone (24 h coverage). The randomization schedule was designed to achieve a 1:1 randomization at each participating center. If the PaO2 decreased to below 50 mmHg for at least 1 h in any patient, all rescue therapies considered appropriate by the principal investigator of the center were allowed (e.g., NO, HFOV, additional surfactant, ECMO, vasodilators). Rotational therapy or prone positioning was obligatory.

The study was conducted as a multicenter, open, randomized parallel comparison. The primary variable was the change from baseline in the PaO2/FIO2 ratio at 48 h after the first administration of surfactant or randomization to the control group. Secondary endpoints were: peak inspiratory pressure, positive end-expiratory pressure, mean airway pressure, in- and expiration time, respiratory rate, FIO2, PaO2, paCO2, SaO2, pH, heart rate, and blood pressure 2, 4, 12, 24, 48, and 120 h after randomization. In addition Murray Score, PRISM III score at baseline, 48 and 120 h, clinical status at 30 days after randomization, days on ventilator, days in intensive care, days on supplemental oxygen, mortality at day 30, ventilator-free days at day 30, and the necessity of rescue therapy as ECMO, HFOV, NO, or rescue surfactant. The primary hypothesis defined was tested by the Mann-Whitney U test at an error level of α≤0.05. A secondary analysis was performed to detect changes from baseline for all other time points up to 48 h by the same test procedure.

For all other variables the same procedures were performed. The number of deaths was compared between groups by means of Fisher's exact test, as were patients who received rescue therapy (each rescue therapy was summarized by frequencies). In a third analysis the combined event death and/or rescue therapy was investigated. We performed multiple regression analyses for changes in the oxygenation index at 2, 4, 12, 24, and 48 h to analyze differences between special subgroups. We used a linear model for repeated measurements. For this post hoc analysis the following subgroups were evaluated: baseline PaO2/FIO2 less than vs. 65 or higher, baseline PRISM III score less than vs. 12 or higher, age under 1 year vs. 2 years or older, girls vs. boys, body weight less than vs. 12 kg or higher, time since FIO2 being higher than 0.5 at randomization shorter than 24 vs. 24 h or longer, time since PIP being higher than 30 cmH2O shorter than 30 h or longer, pneumonia vs. no pneumonia, sepsis vs. no sepsis, and immunosuppression vs. no immunosuppression.

Results

In the pilot study the average increase in PaO2/FIO2 was 54 at 4 h and 103 after 48 h (p<0.01). Seven patients died despite improved oxygenation. The average dose of surfactant given was 94 mg/kg (Table 3). This led to the multicenter study dose of 100 mg/kg.

Table 3. Data of the pilot study in 19 patients, their diagnoses, age, body weight and PaO2/FIO2 ratio at baseline, 4 and 48 h (RSV respiratory syncytial virus)
Full size table

Regarding the primary variable, change in PaO2/FIO2 over the 48-h observation period, PaO2/FIO2 was significantly higher in the surfactant group 2 h after the first surfactant dose (p<0.003). Even after 48 h the surfactant group patients still showed a greater, albeit not significantly greater, increase in PaO2/FIO2 ratio (Fig. 2). Using the Hallman oxygenation index as oxygenation parameter produced similar results; oxygenation was significantly improved after 2 and 4 h (p<0.00222 and p<0.05). Considering secondary endpoints, mortality and mortality and/or rescue therapy was lower in the surfactant group, however not significantly so at all times during the study period. Mortality in both groups was considerably lower than that in the "intended to treat" patients (64%). Patients in the surfactant group received significantly less rescue therapy than those in the nonsurfactant group (p<0.05).

Fig. 2.
figure 2

Medians and interquartile ranges (25–75 percentiles) of the oxygenation index (PaO2/FIO2) in the 48-h observation period of the surfactant group (group 1) and controls (group 2)

Full size image

There was a reduction in mean airway pressures used in the surfactant group after 2 and 24 h (p<0.05, p<0.007; (Fig. 3, Tables 4, 5) – there was no difference between paCO2 or pH in the two groups at any time of the 48-h observation period. Tidal volumes were kept below 10 ml/kg in all cases as specified in Fig. 1. Lung injury and PRISM III scores decreased from 0 to 48 h; however, only the decrease in PRISM III reached the level of significance (p 0.05). No other significant differences in secondary outcome criteria were detected. Eleven patients in the surfactant group received the possible second dose of 100–mg/kg surfactant; no significant increase in PaO2/FIO2 after this second dose was observed at any time. No treatment associated adverse events were observed in the surfactant group; however, the expected risk of intermittent obstruction of the endotracheal tube with a short time deterioration in oxygenation was observed in three patients.

Fig. 3.
figure 3

Medians and interquartile ranges (25–75 percentiles) of changes of peak inspiratory pressure in cmH2O (a) and positive end-expiratory pressure (b) over the 48-h observation period in the surfactant group (group 1) and controls (group 2)

Full size image
Table 4. Physiological and ventilation variables at baseline and after 2, 4, 12, 24, and 48 h (changes from baseline): median (25th/75th percentiles) or mean ±SD (RR ventilatory/respiratory rate, OI oxygenation index, CVP central venous pressure)
Full size table
Table 5. PRISM III and Murray score at baseline, 48, and 120 h (no significant differences)
Full size table

In a post hoc analysis of the PaO2/FIO2 changes from baseline between 2 and 48 h considering various patient characteristics as additional information no significant differences were found regarding Murray Score PRISM III score, age, sex, time with FIO2 longer than 0.5, PIP higher than 30 cmH2O, sepsis vs. no sepsis, or immunosuppression vs. no immunosuppression (Table 6). However, a significant difference in PaO2/FIO2 increase was found between those whose initial ratio was higher than 65 and those whose initial ratio was less and 65 and group 2 patients (p<0.028, Fig. 4); patients with the lower ratio had a 100% mortality in both groups even when rescue therapy was applied. In addition, surfactant patients without pneumonia had significantly better oxygenation after 48 h than nonsurfactant patients with pneumonia (p<0.0017, Fig. 5). In addition, a trend to efficacy of surfactant was seen in patients weighing at least 12 kg (p=0.055).

Table 6. p values in linear models for the change from baseline in the PaO2/FIO2 ratio between 2 and 48 h including various additional factors of clinical concern. The significant interactions between treatment and morbidity characteristics (e.g., pneumonia, PaO2/FIO2<65) indicate that the difference between treatment groups depends on the baseline value of the oxygenation index and on the presence of pneumonia as a causative diagnosis for ARDS. These differences are clearly demonstrated by the median profiles which are displayed separated for the respective subgroups
Full size table
Fig. 4.
figure 4

Z statistic of the primary endpoint, change in PaO2/FIO2 ratio at 48 h between the surfactant group (group 1) and controls (group 2)

Full size image
Fig. 5.
figure 5

PaO2/FIO2 changes from baseline in the group of patients with an initial PaO2/FIO2 ratio less than and greater than 65 (dotted lines) in the surfactant group (group 1) and controls group (group 2)

Full size image

The study had to be stopped earlier than originally planned due to increasing recruitment difficulty. To demonstrate that the trial was not being stopped at a prejudicial time point the primary comparison between treatments was performed sequentially, post hoc beginning with the first ten evaluable patients. The resulting Z statistics, which are approximately normally distributed, are depicted graphically vs. the number of patients (Fig. 4).

Fig. 6.
figure 6

PaO2/FIO2 changes from baseline in the group of patients with (dotted lines) and without pneumonia in the surfactant group (group 1) and controls (group 2)

Full size image

Discussion

In patients with ARDS less endogenous surfactant is produced, and this is inactivated, modified, and not reused, thus causing an absolute and relative surfactant deficiency [20, 24, 25, 26, 23]. As surfactant is a major biological factor for alveolar recruitment, surfactant deficit is a key problem in ARDS, and substitution of surfactant in ARDS could be an important therapeutic tool [19, 23, 26]. As early as 1989 it was hypothesized that surfactant could be of therapeutic value not only in premature infant's respiratory distress syndrome but also in ARDS [24]. In that year the first case report of a successfully surfactant-treated child with severe hypoxemic ARDS was published [27]. Surfactant therapy in preterm infants is now based on sound data (e.g., [28, 29, 30]). In respiratory failure of term infants, in some ways resembling ARDS, a randomized study demonstrated a significant reduction in the need for ECMO [31]. Several uncontrolled studies have been published on adult ARDS [32, 33]; a large controlled study using small doses of aerosolized synthetic surfactant found no advantage of surfactant [33], and a smaller administering a bolus reported improved oxygenation and slightly increased survival in the treatment group [34]. In the pediatric population surfactant has been used in several case reports and in small uncontrolled studies of near fatal ARDS [17, 19, 24, 35]. In mild to moderate ARDS (Hallman oxygenation index less than 10) a randomized study demonstrated improved oxygenation immediately after surfactant administration and most ventilation-associated parameters [36]. In patients with severe hypoxemia we observed a short period of improved oxygenation, as in the study by Willson et al. [36] in patients with less severe hypoxemia.

The originally planned number of patients, however, could not be recruited in the time period scheduled. This might be due to the fact that the overall incidence of ARDS in children decreased dramatically in central Europe during the study period [13, 37]. The study was stopped as the recruitment dropped in this obviously high mortality group. As a sequential analysis of the primary endpoint (normalized Z statistics) demonstrated, the study was not stopped at a point at which there was a significant difference in respect of the primary outcome criteria; this would have been true at an earlier time point.

The overall mortality rate in this study is comparable to that of other studies carried out in severe hypoxemic pediatric ARDS [4, 5, 13]. The trend towards a decreased mortality and need for rescue medication in such a high mortality population might nevertheless be important as no other evidence-based treatment option exists. Surfactant may be only one factor in improving oxygenation in severe ARDS; it enhances the benefits of the "open lung concept," prone positioning, and NO-induced pulmonary vasodilatation [25, 38, 39]. The improvement in oxygenation in the control group, which was managed only on a strictly enforced ventilation algorithm and blood pressure and fluid intake protocol, suggests that standardized treatment alone using conventional methods leads to an overall improvement in patient outcome. This is in line with similar findings by Steinhard et al. [12] who reported improved patient outcome simply by introducing enforced ventilatory management protocols. Surfactant might also be directly involved in improving the balance of pro- and anti-inflammatory mediators in ARDS and therefore be a causative approach [39, 40, 41]. The design of modified surfactant solutions being more prone against inactivation by different proteins in the course of ARDS could enhance and prolong the effect of surfactant on oxygenation [42].

As patients without pneumonia showed a significant improvement and patients weighing more than 12 kg had a benefit from surfactant, it can be speculated that these small infants, in whom pneumonia is the principal cause of ARDS, would require much higher surfactant doses. This would be in accordance with experimental results in which in pneumonia surfactant inactivation was increased, compared with other causes of ARDS [38]. Higher doses of the surfactant preparation used cannot be applied as they would obstruct the major airways. A second dose showed no benefit in our study, either in the data analysis of primary and secondary endpoint criteria or in the post hoc analysis of subgroups. If at all, higher initial doses in selected patients could be considered if a preparation were available with lower volume, lower viscosity, and perhaps more prone to inactivation. Surfactant treatment with a listed price of €200–400/100 mg is still very expensive.

Conclusions

This study reveals the difficulty in obtaining conclusive results from randomized studies in an intensive care setting and patients with a high mortality. Many "intended to treat" patients could not be randomized as they had underlying lung or heart disease, were dying at arrival in the ICU, or had been ventilated for more than 5 days in regional hospitals. An additional difficulty for such a study is the decreasing incidence of ARDS in children. Our study confirmed results of a previous randomized study by Willson et al. [38] reporting improved oxygenation in pediatric patients with mild to moderate ARDS. The improvement in oxygenation was sustained for patients without pneumonia as underlying disease and with a PaO2/FIO2 ratio greater than 65. The latter subgroup does not seem to benefit from any other available rescue tool at this moment. Ventilation variables could be reduced in the surfactant group. The devastating consequences of aggressive ventilation in these children with severe ARDS can probably be ameliorated with surfactant treatment. We conclude that surfactant treatment in severe pediatric ARDS might offer benefits to the patients.