Prostanoids in bronchoalveolar lavage fluid do not predict outcome in congenital diaphragmatic hernia patients

Vasoactive prostanoids may be involved in persistent pulmonary hypertension (PPH) in infants with a congenital diaphragmatic hernia (CDH). We hypothesized that increased levels of prostanoids in bronchoalveolar lavage (BAL) fluid would predict clinical outcome. We measured the concentrations of 6-keto-prostaglandin F1α (6-keto-PGF1α), thromboxane B2 (TxB2), protein, albumin, total cell count, and elastase-α1-proteinase-inhibitor complex in BAL fluid of 18 CDH patients and of 13 control subjects without PPH. We found different concentrations of prostanoids in BAL fluid of CDH patients with PPH: infants with a poor prognosis had either high levels of both 6-keto-PGF1α and TxB2 compared to controls, or high levels of 6-keto-PGF1α only. TxB2 levels showed a large variability in all CDH patients irrespective of outcome. We conclude that prostanoid levels in BAL fluid do not predict clinical outcome in CDH patients.


Introduction
The high mortality of morbidity in children with congenital diaphragmatic hernia (CDH) is largely determined by the severity of lung hypoplasia and persistent pulmonary hypertension (PPH). 1 Despite improved neonatal intensive care, the overall survival rate still does not exceed 60%. 2 Several attempts have been made to predict mortality in these patients. A recent study showed that data reported to the Extracorporeal Life Support Organization Registry of more than 1000 CDH patients treated with extracorporeal membrane oxygenation (ECMO) did not allow discrimination of non-survivors from survivors. 3 In the lung, arachidonic acid metabolites regulate the bronchial and vascular tone, and are involved in in ammatory processes. 4 Increased levels of eicosanoids have been reported in plasma and in bronchoalveolar lavage (BAL) uid of children with PPH who were treated with conventional ventilation or with ECMO. [5][6][7][8] Increased plasma levels of the stable metabolites of the pulmonary broncho-, and vasoconstrictor thromboxane A 2 (TxA 2 ) and the pulmonary vasodilator prostacyclin (PGI 2 ), TxB 2 , and 6-keto-PGF 1a have been observed in CDH patients during episodes of hypoxemia, 9 and in the immediate postoperative period. 10 -12 BAL has been used to evaluate different aspects of in ammatory responses, such as the number of neutrophilic granulocytes and macrophages, albumin, elastase, and a 1 -proteinase inhibitor activity in ventilated preterm infants with respiratory distress syndrome who were likely to develop chronic lung disease. 13 - 15 We hypothesized that in CDH patents, involvement of vasoactive prostanoids in the pathogenesis of PPH would be re ected by increased prostanoid levels in BAL uid, and that these concentrations might have a predictive value with respect to clinical outcome.
The aim of the present study was to measure the concentrations of 6-keto-PGF 1a , TxB 2 , and various in ammation markers in BAL uid of CDH patients who were treated either with conventional ventilation or with ECMO and to evaluate the prostanoid levels in relation to outcome. The levels of different parameters in BAL uid were compared with those of controls without PPH who had similar gestational age and birth weight. 1993 and January 1996. A group of 18 CDH patients were studied; 13 children underwent conventional ventilation (referred to as the CDH-CV group), and ve children were treated with venoarterial ECMO (referred to as the CDH-ECMO group) using standardized treatment protocols. 16 -18 Four CDH patients had been diagnosed prenatally. Ten patients in the CDH-CV group and all ve ECMO patients suffered from respiratory insuf ciency within 6 hours of birth; the other three patients were respiratory insuf cient after 10 hours, 36 hours, and 28 days, respectively. Operative repair by an abdominal approach was performed in 11 conventionally ventilated children and in three ECMO patients after preoperative clinical stabilization; 16 four of the ve patients who died had not been operated on. All 18 patients routinely received antimicrobial prophylaxis. They underwent echocardiography on admission: right-toleft shunting was diagnosed in six children of the CDH-CV group and in all CDH-ECMO patients. 12 Clinical evidence for right-to-left shunting was obtained by preductal and postductal transcutaneous O 2 -saturation differences of . 10% in ve CV-patients and in all ECMO patients.
Thirteen other children without pulmonary abnormalities, who were mainly ventilated perioperatively at a median age of 2.5 days (range 1 ±8 days), served as age-matched controls. They were selected for the best possible match for age, gestational age and birth weight. They all received antimicrobial therapy. Echocardiography was performed in ve controls to exclude structural cardiac anomalies; none of these had evidence of right-to-left shunting.
In all 31 patients cultures of tracheal aspirates were routinely performed every 3 days. The study was conducted according to the principles established in the Declaration of Helsinki and approved by the Medical Ethical Committee of our hospital.

Study design
Of each patient the following data were recorded: gestational age, birth weight, diagnosis, survival, age at admission and at discharge or death, age at start of respiratory insuf ciency, duration of ECMO and/or arti cial ventilation, and duration of supplemental O 2 therapy. Outcome parameters were: mortality and O 2 -dependency at 28 days, as a criterion for chronic lung disease. 19 Bronchoalveolar lavage was performed on day 1, day 3 and day 7 and once every week thereafter as long as endotracheal intubation was continued and the child remained in our Intensive Care Unit. However, an unstable clinical condition was a reason to modify this schedule.
The following data were recorded at each BAL procedure: clinical events such as surgery, mean airway pressure (MAP), oxygenation index (OI), and AaDO 2 . 16 Arterial blood gases were obtained within 6 hours of the time of BAL.

BAL procedure
The procedure was performed directly after routine endotracheal suctioning by the nursing staff using a technique described by Grigg et al. 20 The patient's head turned left, a 5 Fr openended catheter (outer diameter 1.7 mm; Sherwood Medical, Petit Rechain, Belgium) was passed down the endotracheal tube and placed in wedge position. Warmed 0.9% NaCl was instilled in two aliquots of 1 ml/ kg each. Gentle manual suctioning with a 20-ml syringe was directly performed after each aliquot. In most cases the ventilatory circuit was not broken. The whole procedure always took less than 1 min. The recovered uid was immediately processed; 20 ml was aspirated for cell counting, the remaining uid was centrifuged at 900 3 g.
The supernatant was frozen at -808 C, the cells were resuspended in 0.9% NaCl and processed for cytocentrifugation.

Measurements in BAL¯uid
BAL uid was diluted in Tü rk stain (1:10) for cell counting in a Bürker-haemocytometer. Differential cell counts were carried out on cytospin slide preparations with May ±Gruenwald ± Giemsa stain. In the supernatant 6-keto-PGF 1a and TxB 2 were measured by radioimmunoassay as described previously, using standard prostaglandins from Sigma Co. (St Louis, MO, USA) and antibodies from Advanced Magnetics Inc. (Cambridge, MA, USA). 21 Elastase-a 1 -proteinase inhibitor-complex (E-a 1 -PI) was determined using a commercially available kit (PMN Elastase; Merck Immunoassay 11332; Merck, Ger-many). Albumin was measured by immunoprecipitation using N-antiserum against human albumin (Behring OSAL 14/ 15; Behring, France). Human albumin (20%) (CLB, Amsterdam, The Netherlands) was used to obtain standard curves. Total protein was measured as described by Lowry, 22 standard curves were obtained using Preciset 6 g/ 100 ml (Boehringer 125610; Boehringer, Germany).

Data analysis
Data were expressed as median (range) unless stated otherwise. Because of a low yield of BAL uid, not all measurements could be performed for all samples, which resulted in incomplete data. Ventilatory parameters were compared between groups using the non-parametric Mann ±Whitney U-test. Two tailed Spearman's rank correlation coef cient was used to study the relationship between different parameters in BAL uid and ventilatory parameters during the rst 3 days of treatment. In those cases where more than one BAL procedure was performed within the rst 3 days, the average value of each parameter was used for calculation of the correlation coef cient. Statistical signi cance was assumed at the 5%level.

Patient characteristics
The characteristics of all patients are summarized in Table 1. Ventilated controls had the following diagnoses: meconium peritonitis (n = 4), oesophageal atresia (n = 4), M. Hirschsprung, necrotizing enterocolitis, vesical exstrophy, hiatus hernia with Ehlers ±Danlos syndrome, wet lung (each n = 1). None of the controls showed evidence of PPH at any time. Ten CDH-CV patients had left-sided CDH and three had a right-sided defect. Six CDH-CV patients had PPH: two of those, who never met the ECMO entry criteria as a result of a persistently low PaO 2 , died during the rst hours after birth. Four CDH-CV patients were O 2 -dependent at the age of 28 days (one patient without documented PPH). Of the CDH-ECMO patients, who all had documented PPH, one had a bilateral diaphragmatic defect, the other four had left-sided CDH. Two survivors were operated on while undergoing ECMO, and they were both O 2 -dependent at 28 days. The other three ECMO patients died from recurrent episodes of therapy-resistant pulmonary hypertension 5 ±11 days after decanulation; one of these patients was operated on during ECMO on day 20. In these three patients BAL was not performed after decanulation because their clinical condition was unstable and deteriorated rapidly.
In only one CDH-CV patient there was evidence of PPH during the BAL procedure; in all other patients the pre-and postductal transcutaneous O 2 -saturation measurements were similar during BAL.

Prostanoid concentrations in BAL uid of CDH patients
In all but one patient the prostanoid concentration in BAL uid could be measured at least  once; the sample volume of one CDH-CV patient, who died within several hours after birth, was too small to measure prostanoid concentrations. The concentrations of 6-keto-PGF 1a and TxB 2 are shown in Figs 1 and 2, respectively and in Table 2. Once CDH-CV patient with evidence of PPH at the time of BAL who died during the rst day of life had a very high level of 6-keto-PGF 1a , the same was true for the initial levels of two ECMO-treated CDH patients who died later (Fig. 1A). In the group of CDH survivors, only one patient who needed arti cial ventilation for 4 weeks showed high levels of 6-keto-PGF 1a , whereas all other CDH-CV and CDH-ECMO survivors had 6-keto-PGF 1a levels within the control ranges (Fig. 1B). The concentrations of TxB 2 varied irrespective of outcome (Fig. 2). Deterioration of the clinical condition in one patient with PPH who developed sepsis on day 10, followed by thrombosis of the inferior vena cava, was not re ected by an increase in prostanoids initially, but by an increase in TxB 2 a few days later; 6keto-PGF 1a remained low.
Surprisingly, in two infants who died, one CDH-CV patient and one CDH-ECMO patient,  the ratio of 6-keto-PGF 1a to TxB 2 was high: 20 and 5.2 respectively. The median ratio of 6-keto-PGF 1a to TxB 2 of the CDH-CV patients with PPH did not differ from that of those without PPH: 0.52 (0.14 ±20) versus 0.51 (0.02 ±2.5), respectively; controls without PPH had a median ratio of 0.6 (0.14 ±1.6). In the ve CDH-ECMO patients, who all had PPH, the ratio of 6keto-PGF 1a to TxB 2 was 0.7 (0.23 ±5.2). The prostanoid concentrations did not correlate with any of the ventilatory parameters in the conventionally ventilated groups.

In¯ammatory markers in BAL¯uid and correlation between different parameters
Cultures of tracheal aspirates were negative in all patients at the time of BAL. The total cell count, the percentage of neutrophilic granulocytes and macrophages, and the concentrations of protein, albumin, and E-a 1 -PI in BAL are shown in Table 3. During the rst days the median cell count and the levels of protein, albumin, and E-a 1 -PI were high in the CDH-ECMO group.

Discussion
We have found different concentrations of prostanoids in BAL uid of CDH patients with PPH: in infants who died and in one infant ventilated for more than 4 weeks either high levels of 6-keto-PGF 1a and TxB 2 compared to controls, or high levels of 6-keto-PGF 1a only. The ratio of 6-keto-PGF 1a to TxB 2 was high in two CDH patients who died. Survivors with evidence of PPH treated with ECMO, and CDH patients with O 2 -dependency at 28 days had prostanoid levels which could not be discriminated from the levels of the other CDH patients. The levels of TxB 2 were variable in all CDH patients. Increased levels of TxB 2 and 6-keto-PGF 1a have been reported in plasma of neonates with PPH who were treated with conventional ventilation or with ECMO. 6,7,9 -12 Dobyns et al.  described increased levels of TxB 2 , 6-keto-PGF 1a , PGD 2 , PGE 2 , LTB 4 and LTE 4 in BAL uid of neonates with PPH. 8 Prostanoid levels in plasma and in BAL uid were shown to decrease during the course of treatment in these children 7,8 and in one CDH patient treated with ECMO. 11 Dobyns et al. found persisting high levels of 6-keto-PGF 1a and TxB 2 in BAL uid of ECMO-treated children with PPH and a poor outcome, whereas the prostanoid levels decreased rapidly in PPH patients with a good outcome. 8 Our ndings of high levels of 6-keto-PGF 1a in infants who died are in accordance with these ndings, but the variable TxB 2 levels in all CDH patients ± irrespective their clinical outcome ± contradict their observations.
To determine whether the high levels of prostanoids in BAL uid of some CDH patients with PPH are a speci c feature of the abnormal pulmonary vasculature in CDH, these levels should be compared with prostanoid levels in BAL uid of children without CDH who need ventilatory support to the same extent as the most severely ill CDH patients. Our control population consisted of infants who had mild ventilatory requirements as re ected by the ventilatory parameters. Therefore, these controls did not allow for such comparison.
The relatively high and variable TxB 2 levels in the CDH patients, irrespective of the outcome, suggest that the clinical situation in this group of patients is not adequately re ected by the TxB 2 concentrations in BAL uid. An earlier study from our group showed a correlation between plasma prostanoid levels and ventilatory parameters in CDH patients. 12 We were not able to demonstrate such correlations in BAL in our study. This may indicate that prostanoid levels in the pulmonary vasculature are not adequately re ected in BAL uid, as has been suggested by Abman et al. 23 This was supported by our nding that in non-ventilated neonatal rats with CDH the concentration of TxB 2 is 10fold higher in lung tissue than in BAL uid. 24 However, in these rat pups the ratio of 6-keto-PGF 1a to TxB 2 was consistently increased in lung tissue and in BAL uid compared with controls directly after birth, suggesting that this parameter in BAL uid may be representative for the values in lung tissue. 24 In the present study a similar observation was made in two CDH patients, but this was not a consistent nding for all patients with a poor outcome. The increased ratios of 6-keto-PGF 1a to TxB 2 suggest that the lungs compensate maximally for the pulmonary vasoconstriction in CDH. It is noteworthy that in neonatal rats with CDH the presence of more pulmonary neuroendo- Table 3. Total cell count, cell differentiation, albumin, total protein and elastase-a 1 -proteinase inhibitor complex (E-a 1 -PI) in BAL¯uid of ventilated controls and CDH patients, and in CDH patients treated with ECMO  28 We speculate that in some CDH patients the lungs already compensate maximally for pulmonary vasoconstriction and may be insensitive to therapeutic interventions with pulmonary vasodilators, such as prostacyclin and nitric oxide. Increased cell counts, neutrophil numbers, albumin, and elastase activity have been reported in tracheal aspirates or BAL uid of prematurely born ventilated patients who developed chronic lung disease. 13 -15 in the present study, concentrations of cells, protein, albumin, and E-a 1 -PI complex in the BAL uid of CDH-ECMO patients were high, compared with conventionally ventilated CDH patients and controls during the rst days of treatment. To our knowledge these parameters have not been reported before in lung lavages of ECMO-treated neonates. We assume that increased vascular permeability with an in ux of cells is responsible and related to lung injury and activation of the in ammatory cascade before ECMO, or to neutrophil activation and protein in ux as part of the capillary leakage syndrome in the initial phase of ECMO. 29 Our study does not allow for de nite conclusions regarding the question of whether the high cell counts, and the high levels of protein, albumin, and elastase-a 1 -PI-complex in ECMOtreated CDH patients result from the disease state or from the ECMO procedure.
In conclusion, we found that in some CDH patients with PPH high prostanoid concentrations in BAL uid were associated with a poor outcome. These patients died within a few hours of birth, or from recurrent episodes of therapy-resistant pulmonary hypertension several days after ECMO. The high levels of 6keto-PGF 1a might have been induced by the hypoxic vasoconstriction in these patients 30 or re ect an attempt by the lungs to compensate maximally for the pulmonary vasoconstriction. The variation in TxB 2 concentrations may reect lung injury and increased vascular permeability. This study illustrates the dif culties of performing BAL procedures on a regular basis in severely ill neonates and interpreting the prostanoid levels in BAL uid. Based on our ndings we feel that the predictive value of prostanoid levels in BAL uid with respect to outcome is questionable.