INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Analysis of fluid washed from the lung by bronchoalveolar lavage (BAL) gives valuable insight into the composition of the epithelial lining fluid (ELF) covering the alveolar surface. Constituents of BAL fluid have been extensively studied by both clinician and researcher in the diagnosis and investigation of lung disease (1). BAL performed during flexible bronchoscopy has become a routine procedure for collection of a sample of ELF in both the adult (1) and the pediatric population (2, 3), including subjects requiring mechanical ventilatory support because of severe lung disease (4). However, in ventilated neonates and young infants, bronchoscopic BAL is rarely practical. The smaller pediatric bronchoscopes (2.2 mm and less) have no suction port; steerable bronchoscopes with a suction channel (3.5 mm and larger) cannot be passed down infant endotracheal tubes, therefore necessitating extubation to perform the procedure.

In ventilated neonates, particularly preterm neonates, lavage fluid for analysis of ELF constituents has generally been collected by tracheal aspirate (TA), using small aliquots of saline of variable number and volume, instilled down the endotracheal tube and removed by tracheal suction (5). Recently, nonbronchoscopic methods of BAL (NB-BAL) have been described in ventilated infants and children (11, 12), and such techniques have been performed in neonates, including premature infants weighing less than 1,500 g (13, 14). NB-BAL differs from TA in that larger aliquots of saline are used, the suction catheter is advanced further into the airway, and the lavage is conducted through the catheter rather than directly down the endotracheal tube.

When planning a study examining ELF surfactant concentration in neonates and infants with respiratory failure, the findings of which have been published in part elsewhere (15, 16), we encountered a deficiency of data directly comparing TA and NB-BAL for collection of lung lavage specimens. A single study in eight tracheotomized rabbits found that 0.5 to 1 ml TA and 5 ml/kg BAL gave similar mean values for ELF surfactant concentration, but did not report whether the two techniques gave similar estimates in individual animals (6). Additionally, there have been very few attempts to formally evaluate whether small volume lung lavage is truly indicative of ELF composition. We therefore included in the study a sampling protocol to allow direct comparison of two methods of small volume lung lavage, TA and NB-BAL, for collection of fluid for surfactant analysis. The main aims of this endeavor were: (1) to investigate the apparent origin of the return fluid from TA and NB-BAL by comparing the concentration of proteins secreted proximally and distally in the tracheobronchial tree; (2) to examine whether, if performed contemporaneously, the two methods would give equivalent estimates of key surfactant indices, with agreement in individual sample pairs; and (3) to compare the relative accuracy of TA and NB-BAL sampling by determining which method gave estimates of surfactant indices that best predicted alveolar surfactant status, as indicated by severity of lung dysfunction. Additionally, we wished to document the relative safety of the two lavage methods in our patient population.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient Selection

All neonates and infants admitted to our institution with respiratory disease were screened for potential eligibility for this study. Enrollment criteria were: (1) intubated and mechanically ventilated because of respiratory failure; (2) lung disease which was diffuse and bilateral, but not necessarily homogeneous (neonates with congenital diaphragmatic hernia were thus unsuitable); and (3) relatively stable indices of cardiorespiratory function, not requiring major alterations in ventilator settings or other supports at the time of sampling. Informed parental consent was obtained, and the study protocol was approved by our institutional Ethics in Human Research Committee.

Sampling Procedure

TA and NB-BAL samples were collected in each infant by a single operator (P.A.D.). Data regarding cardiorespiratory status, ventilator settings, and most recent arterial blood gas analysis were first recorded, and oxygenation index (OI) and alveolar-arterial oxygen difference (AaDO₂) were calculated.

The plasma urea concentration taken closest in time to the lavage was noted. The lavage sequence was performed at least 1 h after the last episode of routine suctioning of the endotracheal tube.

The two methods of lavage were conducted sequentially, with an interval of approximately 5 min separating the two. TA sampling was done first in each case, as this method involves a smaller lavage volume delivered more proximally, and was thus felt less likely to interfere with the subsequent technique (NB-BAL) than vice versa. Prior to lavage, peak inspiratory pressure was increased by 2 cm H₂O, and fraction of inspired oxygen (FI_O2) by 0.1 to 0.2; hand ventilation with an anesthetic bag was used in infants with more severe lung disease, and/or those on high-frequency oscillation. Further alterations were made during the procedure as necessary; after lavage, ventilation settings were returned to baseline over a 10-min period.

The TA method chosen for the study has been frequently used for collection of lavage fluid specimens in ventilated newborn infants (7- 9). A standard suction catheter was measured so that, upon suctioning, the catheter tip would be positioned in the lower trachea, 1 to 2 cm beyond the end of the endotracheal tube. With the infant lying supine, and head in the midline, 0.5 ml 0.9% saline was instilled down the endotracheal tube via a side-port connector (Portex Ltd, Hythe, UK). Ventilation was then reestablished for five breaths, the circuit disconnected, and the suction catheter passed into the trachea. Suction was applied using 200 mm Hg negative pressure, and the lavage fluid was collected in a mucus trap. This procedure was repeated for a total of four aliquots of saline, taking approximately 2 min to complete. The catheter was then rinsed with 1 ml of saline, and the specimen refrigerated before further processing and analysis as described subsequently.

NB-BAL was initially conducted exactly as described by Alpert and coworkers (11), using balloon wedge-pressure catheters (No. AI 07121; Arrow Inc., Reading, PA). In view of a high proportion of significantly blood-stained specimens, an alternative method was adopted (12, 13) using a straight, snub-nosed, end-hole suction catheter (No. 565; Vygon, Ecouen, France). The infant was positioned supine with the head turned 90° to the left; such a position virtually ensures that a catheter advanced down the trachea will enter the right main bronchus (17). While ventilation continued, the catheter was introduced into the endotracheal tube by way of a suction bullet in a swivel Y-connector, and gently advanced until resistance was met. A size 8 suction catheter was used for endotracheal tube size 3.5 mm internal diameter, size 6 for 3.0 mm or smaller. An aliquot of 1 ml per kg body weight of saline, warmed to 37° C, was instilled through the catheter, coinciding with a positive pressure inflation of the lung. The dead space within the catheter was cleared with air after each injection. Low-pressure suction was then immediately applied (60 to 90 mm Hg for a size 8 catheter, 80 to 120 mm Hg for a size 6), and the return fluid collected in a mucus trap (Davol Inc., Cranston, RI). A total of three aliquots were thus instilled, with suction after each; total duration of the lavage was approximately 60 to 90 s. The fluid returned from all aliquots was pooled, and processed as described subsequently.

Paired samples were taken using the two sampling techniques as soon as was practical after enrollment in the study. Sequential paired samples one or several days apart were taken in many of the infants while still ventilated.

Sample Processing and Analysis

All specimens were centrifuged at 150 × g and 4° C for 10 min; aliquots of lavage supernatant were stored at 20° C for later analysis. The presence of any mucus and debris in the supernatant after centrifugation was noted. Blood staining of the specimen was quantified as follows: 0 = no blood staining; 1 = no visible blood staining of lavage fluid, slight blood contamination of the pellet after centrifugation; 2 = no visible blood staining, red blood cell layer comprising up to half the total volume of the pellet; 3 = barely perceptible blood staining of fluid, red blood cell layer comprising more than half the total volume of the centrifugation pellet. Samples that were heavily contaminated by mucus, or showed significant macroscopic blood staining, were discarded.

Total surfactant phospholipid (PL) in the lavage fluid was measured by the method of Bartlett (18), and disaturated phosphatidylcholine (DSPC) was assayed by alumina column chromatography (19). Concentration of surfactant protein A (SP-A) was measured by enzyme immunoassay (20), and of sphingomyelin (SM) by high-performance liquid chromatography, using a Diol column (E. Merck, Darmstadt, Germany), as described by Andrews (21). Total protein concentration was assayed using a micromodification of the method of Smith and coworkers (22), with bicinchoninic acid as the chromogen (BCA Protein Assay Reagent; Pierce Chemical Co., Rockford, IL). Lipid turbidity in the sample was eliminated by addition of sodium dodecyl sulfate (final concentration 0.5%) and Triton-X 100 (final concentration 0.5%); neither of these agents at these concentrations interferes with the assay (22). IgA secretory component (SC) was measured with an enzyme immunoassay as previously described (15), using two commercially available antibodies to SC. Purified secretory IgA was used as the standard, and thus sample concentrations of free and bound SC were expressed as µg/ml equivalent of secretory IgA standard. Urea concentration in lavage fluid was measured by urease assay, and albumin by immunoturbidimetry, both using an automated analyzer.

The amount of ELF in the lavage fluid was calculated (23), according to the formula: ELF volume = ([urea]_LAVAGE/[urea]_PLASMA) × return fluid volume. The ELF concentration of all analytes was calculated (e.g., [SP-A]_ELF), as were the following ratios: protein, SC and SP-A to albumin, DSPC to SM (i.e., the L/S ratio), and DSPC to PL.

Data Analysis

Logarithmic transformation of all data sets pertaining to lavage fluid indices was undertaken, other than DSPC/PL ratio, which was consistently normally distributed. Means or geometric means were compared by paired t test. Sampling precision and bias were investigated for the major surfactant indices by plotting the difference between the TA and NB-BAL values against the mean of the two (i.e., the best estimate of the true value) (24). For the log transformed indices, precision of the NB-BAL estimate was expressed numerically as a ratio to the TA value after back transformation of the 95% confidence intervals (CI) for the spread of the data points on the y-axis (24). Bias was assessed by examining whether the mean of the differences differed significantly from zero. For each surfactant index, multiple linear regression of the paired TA and NB-BAL values (i.e., two predictor variables) was performed to determine which estimate best predicted OI (the outcome variable).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient Data

A total of 65 paired samples of lung lavage fluid were collected in 36 infants with respiratory failure requiring ventilatory support. Demographic data for these infants are shown in Table 1, and data related to sample collection in Table 2. In 12 sample pairs, one of the samples was unsuitable for analysis because of contamination by mucus (3 TA, 0 NB-BAL samples), macroscopic blood staining (2, 6), or inadequate sample volume (0, 1). Full sample analysis was thus able to be performed in 53 paired samples (from 31 infants), upon which the main findings of this study are based. Where appropriate, subgroup analysis was performed on the first sample pairs obtained in each infant (n = 31), and on a group of samples taken preextubation in 14 infants (designated as "last sample"). Fourteen of the 65 NB-BAL samples were collected using a balloon-tip catheter (11), of which four were discarded because of macroscopic blood staining. Exclusion of these samples from the data set did not alter the findings in any significant way.

Sampling Data

The degree of desaturation, and the time for recovery, were similar for the TA and NB-BAL lavage procedures (Table 2). OI and AaDO₂ after the paired sampling were not significantly altered from prelavage values (mean change in OI 0.18, p = 0.74, paired t test; change in AaDO₂ 9.7, p = 0.28). Severity of lung disease (OI or AaDO₂) did not predict whether desaturation would occur during a lavage procedure, nor did these indices correlate with recovery time after the lavage. Particularly for NB-BAL, no significant linear association was noted between percentage lavage return and degree of desaturation during sampling (Spearman's r = 0.11, p = 0.45), or increase in OI postsampling (r = 0.083, p = 0.56). Only two patients had desaturation episodes with lavage lasting longer than 5 min; the first was a 3-month-old ex-preterm infant with viral pneumonitis, in whom desaturation occurred for 7 min during both TA and NB-BAL sampling, and the second was a 9-d-old term infant with hydrops fetalis, in whom desaturation occurred for 10 min during NB-BAL. In neither case was there any deterioration in OI and AaDO₂ at the time of the next arterial blood gas analysis.

In three study infants with meconium aspiration, NB-BAL appeared to have a beneficial effect, leading to a reduction in AaDO₂ after the procedure (76, 95, and 98 mm Hg), in association with the appearance of large amounts of meconium debris in the lavage pellet. This finding is in keeping with previous reports of the efficacy of saline lavage in selected infants with meconium aspiration (25).

The mean distance of wedging of the NB-BAL catheter beyond the endotracheal tube tip was 5.3 cm (range 4 to 7 cm). This appeared not to be affected by the diameter of the catheter.

Return Fluid Characteristics

Proportion of lavage fluid recovered was equivalent for TA and NB-BAL sampling, but a greater percentage of the NB-BAL return fluid was recoverable as supernatant after centrifugation at 150 × g (Table 2). TA supernatant was more likely to contain mucus and debris, whereas NB-BAL samples were more frequently contaminated with blood. Blood staining was a consistent feature in NB-BAL specimens from infants with meconium aspiration syndrome (all 12 samples scored  2), but infrequent in viral bronchiolitis (1 of 15 scored  2). Additionally, in meconium aspiration the centrifugation pellet frequently consisted predominantly of meconium debris, with variable degrees of opalescence of the supernatant. A suspected clinical diagnosis of meconium aspiration was confirmed in several infants on the basis of heavy meconium contamination of lavage fluid.

Protein Concentrations and Ratios

Table 3 shows the raw concentration of protein, SP-A, and SC in lavage fluid, the calculated ELF concentration, and ratio of each component to albumin. Compared with NB-BAL, TA fluid had a higher protein content, and specifically a higher concentration of SC, a product of the respiratory epithelium secreted proximally in the tracheobronchial tree (26). The ratio of SC to albumin was fivefold higher in the TA samples. On the other hand, ELF concentration of SP-A (secreted in the terminal bronchioles and alveoli [27]), as well as the SP-A/ albumin ratio, were higher in NB-BAL samples.

Equivalence and Agreement of Surfactant Index Measurements in Paired Samples

Plots of difference versus mean for four key surfactant indices measured in TA and NB-BAL samples are shown in Figure 1, with mean values, interpretation of the confidence limits, and mean of the differences quoted in the legend. For each index, the mean of all values was equivalent in TA and NB-BAL samples, other than for [SP-A]_ELF, where a lower mean value was noted in TA specimens. There was no suggestion of a consistent bias relating to the standardized order in which the samples were taken. For all indices, the individual TA and NB-BAL samples show relatively poor agreement, as indicated by the widely disparate 5% and 95% confidence limits on the Bland-Altman plot. No clear change in the precision of measurement was seen in analysis of the first and last sample subgroups; the only additional example of bias was in the last sample group, where there was a trend toward higher [DSPC]_ELF values in TA samples compared with NB-BAL (1.10 versus 0.63 mg/ml; p = 0.059).

View larger version (32K): [in this window] [in a new window]   Figure 1.   Precision and bias analysis in paired TA and NB-BAL samples. Plots difference (TA  NB-BAL) versus mean after log transformation where appropriate. (Panel A) DSPC/PL ratio (raw data). Equivalence: TA mean 0.39; NB-BAL 0.38. Precision: NB-BAL value may be 0.27 below, to 0.30 above, the TA value. Bias: Not detected. (Panel B) DSPC/SM ratio (log transformed). Equivalence: TA geometric mean 22. NB-BAL 22. Precision: NB-BAL value lies between 0.22 and 4.4 times the TA value. Bias: Not detected. (Panel C ) [DSPC]ELF (log transformed). Equivalence: TA geometric mean 0.62 mg/ml. NB-BAL 0.50 mg/ml. Precision: NB-BAL value lies between 0.082 and 19 times the TA value. Bias: Trend toward higher [DSPC]ELF in TA samples, not statistically significant (p = 0.24). (Panel D) [SP-A]ELF (log transformed). Equivalence: TA geometric mean 9.8 µg/ml; NB-BAL 19 µg/ml. Precision: NB-BAL value lies between 0.015 and 17 times the TA value. Bias: Mean of differences (ln values) 0.68; (95% CI 1.16, 0.20). Significantly lower [SP-A]ELF values in TA samples (p = 0.0065).

Prediction of OI

TA and NB-BAL estimates of [DSPC]_ELF, DSPC/SM, and [SP-A]_ELF showed a negative association with OI, with multiple correlation coefficients of 0.43, 0.43, and 0.46 respectively, in multiple linear regression analysis. Both for [DSPC]_ELF and [SP-A]_ELF, the TA estimate was independently predictive of OI in the linear regression equation (p = 0.03 and 0.046 respectively), whereas the NB-BAL estimate was not. For DSPC/ SM, neither estimate showed significant independent association with OI. Using AaDO₂ as the outcome variable in the regression equations produced a similar result. The association of surfactant indices with OI and AaDO₂ did not change from first to last sample.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Comparison of Quantitative Analysis of TA and NB-BAL Samples

In this study, we firstly sought to compare the results of quantitative analysis of fluid collected by contemporaneous TA and NB-BAL. The TA and NB-BAL protocols used were representative of those previously described, but particularly for TA there is wide variation in reported aliquot volumes (ranging from 0 to 2 ml), and aliquot number (range 1 to 5). We used four aliquots of 0.5 ml saline for TA as this protocol has been relatively widely employed in previous studies involving collection of lung fluid from ventilated infants (7), including investigations of pulmonary surfactant concentration (7). Our initial NB-BAL protocol, following the method of Alpert and coworkers (11), resulted in relatively low fluid return volume, and a high proportion of heavily blood-stained aspirates. We therefore changed to a simpler NB-BAL method incorporating elements of two previous reports (12, 13). In contrast to Koumbourlis and Kurland (12), the NB-BAL catheter we used did not have side-holes, thus facilitating delivery of the instilled saline aliquots to the distal airspaces beyond the point of wedging. Three aliquots of 1 ml/kg saline were chosen for this study both to ensure that the distal airspaces were sampled (13), and to conform with the lavage protocol relatively widely used for BAL during pediatric fiberoptic bronchoscopy (2, 3).

Comparison of protein concentrations in the paired specimens indicates that the two lavage methods are sampling different parts of the tracheobronchial tree. The higher total protein content in TA fluid, whether expressed as a raw concentration, an ELF concentration, or as a ratio to albumin, clearly suggests that the return fluid has a different origin in each case. The higher SC concentration in TA samples is indicative of the TA lavage washing the airway mucosa more proximally than NB-BAL. Secretion of SC in the airway is centered in the larger conducting airways, and does not occur in terminal bronchioles or alveoli (26). Rennard and coworkers (28) showed that the ratio of IgA to albumin was significantly higher in the first (bronchial) return fluid from fiberoptic BAL than in subsequent lavage aliquots. A difference in this ratio was not noted by Merrill and coworkers (29), but higher aliquot volume (50 ml versus 20 ml) may have resulted in the return fluid from the first lavage aliquot having a more distal origin. Their finding of an inverse correlation between secretory IgA/albumin ratio (or SC/albumin ratio) and fluid recovery in the first lavage aliquot supports the notion that selective lavage of the proximal airways will result in relatively higher SC concentrations in the return fluid.

The lesser SP-A concentration in TA fluid provides further evidence of its proximal origin compared with NB-BAL. Although some SP-A secretion may occur in large airway epithelium (30), production of significant amounts of SP-A appears to be essentially confined to the epithelium of the alveoli and smallest airways (27). The difference in SP-A concentration in our study must therefore represent a more distal sampling of the airspaces during NB-BAL than TA. Raw concentrations of SP-A were not different in TA and NB-BAL samples, but did show extreme variability, highlighting the need for a dilutional marker against which to reference the concentration of analytes measured in lavage fluid (see accompanying article on pp. 778-784).

The differing origins of the return fluid from the two sampling techniques is a reflection of the important methodological differences between TA and NB-BAL. The use of an end-hole only catheter, wedging of the catheter in a large bronchus, and the instillation of larger fluid volumes (relative to TA) all predict a more distal lavage during NB-BAL. In this study, the NB-BAL catheter wedged on average 5.3 cm beyond the endotracheal tube tip. On the basis of previous radiologic studies (13, 17), and the pathologic data of Horsfield and coworkers (31), we would assume that the point of wedging will be within a lobar or segmental bronchus in the right lung. Fluid instilled down a catheter wedged in this position appears, at least radiographically, to migrate distally in the tracheobronchial tree (13), although this information was obtained post mortem in a single infant with relatively deaerated lungs. By contrast, in adult subjects, the first aliquot of radiopaque fluid (approximate volume 1 ml/kg) instilled at fiberoptic bronchoscopy appeared to remain in the central airways, with limited dispersion into the peripheries of the lung (32). The apparent difference in distribution of a single 1 ml/kg aliquot of saline in these two studies may be in part methodological, but will also be a function of the volume of the conducting airways relative to body size. Tracheal, as distinct from bronchial, fluid instillation would appear to result in less distal migration down the tracheobronchial tree. Tracheal instillation of radiolabeled saline in dogs and adult humans resulted in minimal dispersion of radioactivity beyond the mainstem bronchi (33). There are no corresponding pediatric data available, and the applicability of this finding to neonates and infants may be limited.

In our study population, the mean values for key surfactant indices were similar in the TA and NB-BAL samples (other than for [SP-A]_ELF, see Figure 1). This suggests that, excepting SP-A, estimates of surfactant indices in a group of ventilated infants will be equivalent with either method of small volume lavage. This finding is in keeping with a previous comparison of TA (0.5 to 1 ml) and BAL (single 5-ml bolus) in tracheotomized rabbits (6), where similar values for [PC]_ELF and [DSPC]_ELF were obtained with the two lavage methods. No data were presented in this report on agreement between paired estimates in individual animals. In the present study, the potential imprecision of a single episode of small volume lavage in an individual subject was highlighted by the results of the Bland-Altman analysis of difference versus mean for the paired samples. For all surfactant indices, there was little agreement between the individual TA and NB-BAL values, some of which differed by several orders of magnitude. It is thus likely that a single episode of lung lavage may at times yield a specimen in which measured surfactant concentration is not at all indicative of true ELF concentration. This imprecision will be amplified where samples are taken by multiple operators, using a slightly different lavage technique on each occasion. As has been suggested previously (34), a highly standardized lavage protocol is mandatory in any study involving analysis of lung lavage fluid; even with such a protocol, results from individual patients should be interpreted with caution.

Prediction of Lung Dysfunction in the Paired Samples

In order to compare the accuracy of estimates of surfactant indices in the TA and NB-BAL samples, in this investigation we chose two measures of gas exchange, OI and AaDO₂, as indirect markers of alveolar surfactant concentration. Clearly these physiological parameters will be affected by many other factors in our heterogeneous patient population; the correlation coefficient values suggest that approximately 20% of the variability in OI is explained by alteration in surfactant indices. We consider that this represents a sufficient association to allow a meaningful comparison of the TA and NB-BAL estimates of surfactant indices to be made, if only to determine whether one is clearly superior to the other in prediction of OI. The somewhat surprising result was that TA samples were at least equal to the NB-BAL samples in prediction of lung dysfunction. This suggests that, even in lavage samples from the proximal airway, the concentration of alveolar constituents in the return fluid is to some extent indicative of their concentration in alveolar ELF. Return fluid from dry suction of the trachea or pharynx soon after birth is certainly representative of alveolar fluid composition (35), presumably related to efflux of lung fluid via the airway after delivery. Some clearance of alveolar constituents up the trachea is maintained beyond the newborn period (36), and TA specimens from the adult porcine lung contain surfactant that is of alveolar origin (37). These findings most likely explain why TA sampling, even without saline lavage (38), yields material that is indicative of alveolar fluid composition. This efflux may also permit accurate quantitation of ELF constituents other than surfactant in lavage fluid, but the validity of such measurement cannot readily be determined from our study data alone.

Safety of Small Volume Lavage

The data collected during each sampling episode in this study confirm that TA and NB-BAL can be conducted safely in a population of infants with significant lung disease. Both methods were associated with transient arterial desaturation, but did not result in any prolonged compromise of gas exchange. We did not demonstrate a propensity to desaturation based on severity of lung disease, or volume of fluid returned. During NB-BAL, 42% of instilled saline was recovered, implying that after instillation of three aliquots of 1 ml/kg, on average 1.7 ml/kg remained in the lung (i.e., 6% of normal infant FRC). The absence of a correlation between return fluid volume and degree of desaturation, or gas exchange at subsequent arterial blood gas analysis, suggests that the retained fluid after NB-BAL rapidly disperses, presumably by incorporation into the ELF, and flux into the pulmonary interstitium and capillaries (39).

In conclusion, this study has shown that, compared with NB-BAL, TA return fluid from the ventilated infant is more proximal in its origin, but yields estimates of surfactant indices that are predictive of lung dysfunction. Individual measures of surfactant concentration show poor agreement in contemporaneous paired TA and NB-BAL samples, highlighting the potential imprecision associated with a single episode of small volume lung lavage. Both methods of small volume lavage are well tolerated even in infants with significant lung disease, but preoxygenation and adjustment of ventilation settings are necessary to maintain cardiorespiratory stability.