Published ahead of print on September 25, 2003, doi:10.1164/rccm.200307-920OC
© 2003 American Thoracic Society A Simple Method for Estimating Respiratory Solute Dilution in Exhaled Breath CondensatesDivision of Pulmonary and Critical Care Medicine, Gastroenterology, and Adult Cystic Fibrosis Center, Medical College of Wisconsin, Milwaukee; and Respiratory and Critical Care Division, Zablocki V.A. Hospital, West Allis, Wisconsin Correspondence and requests for reprints should be addressed to Richard M. Effros, M.D., 30404 Camino Porvenir, Rancho Palos Verdes, CA 90275. E-mail: effros{at}mcw.edu
Exhaled breath condensates have been widely used to detect inflammatory mediators in the fluid that covers airway surfaces of patients with inflammatory lung disorders. This approach is much less invasive than bronchoalveolar lavage, but respiratory droplets are markedly diluted by large and variable amounts of water vapor. We estimated the dilution of respiratory droplets by comparing concentrations of nonvolatile, reference indicators (total nonvolatile cations, urea or conductivity) in 18 normal subjects with normal plasma concentrations by assuming similar concentrations in the respiratory fluid and plasma. The volatile cation, NH4+ (most of which is delivered as NH3 gas from the mouth), represented 93 ± 3% (SEM) of the condensate cations. More than 99% of the NH4+ was removed by lyophilization, making it possible to use conductivity to estimate total nonvolatile ionic concentrations and facilitating analysis of urea. Conductivity was significantly correlated with electrolyte and urea concentrations. Estimates of dilution based on total cations, conductivity, and urea were not significantly different (cations: 20,472 ± 2,516; conductivity: 21,019 ± 2,427; and urea: 18,818 ± 2,402). These observations suggest that the conductivity of lyophilized samples can be used as an inexpensive, simple, and reliable method for estimating dilution of nonvolatile, hydrophilic mediators in condensates.
Key Words: freeze drying • electrical conductivity • cations • urea The broad appeal of exhaled breath condensate studies is readily understandable (1, 2). Until recently, bronchoalveolar lavage has been the only way that fluid lining the airways and airspaces could be sampled in patients who do not produce sputum. Although bronchoalveolar lavage can yield important clinical and investigative information, it involves some risk, including infection, impairment of gas exchange, and problems with sedation. Furthermore, bronchoalveolar lavage is expensive and inconvenient, cannot be readily repeated at frequent intervals, and entails a number of artifacts associated with the instillation of fluid into the airways (3). In contrast, collection of condensates is noninvasive, can be repeated as often as needed, and does not alter the fluids on the lung surfaces. Unfortunately, the condensate method involves some rather formidable problems of its own. Perhaps the most serious difficulty associated with exhaled condensates is that of extreme and what may be variable dilution of respiratory droplets (4). Most of the water that is exhaled from the lungs is generated in the form of water vapor, a gas that cannot carry nonvolatile solutes. The formation of droplets from exhaled water vapor can be considered as an artifact associated with cooling. The quantity of water vapor generated each minute by the lungs is determined by the ventilatory rate and the saturation of the exhaled air, which is close to 100%. Full saturation of the exhaled air keeps the airway surfaces moist, thereby enhancing gas exchange. Water exchange is presumably assisted by the presence of a variety of aquaporins on airway surfaces (5). In contrast, little is known about release of droplets from the respiratory surfaces. We recently drew attention to the need for dilutional reference indicators that would permit calculation of solute concentrations in the respiratory fluid from those in the condensates (4). Although increases in the concentrations of inflammatory mediators that have been described in a variety of lung diseases could reflect increases in the concentrations of these indicators in respiratory fluid, they could also be due to an increase in the volume of respiratory droplets that reaches the condenser and is then dissolved in the water vapor that has collected there. There is reason for believing that changes in condensate concentrations of inflammatory mediators in lung disease are not simply due to alterations in dilution. Comparable increases in the concentrations of these mediators have been observed when bronchoalveolar lavage samples have been collected in these patients. Furthermore, increases in the concentrations of some markers of inflammation were associated with no change or even decreases in the concentrations of other indicators (6–9). However, measurements of relative concentrations of inflammatory markers may be misleading. An increase in the relative concentration of one mediator to that of a second mediator could be due to an increase in the first or a decrease in the second. Rather than using another inflammatory mediator to determine whether changes in the dilution of respiratory droplets had occurred, it would be preferable to use noninflammatory reference indicators that remain relatively unchanged in the respiratory fluid and, ideally, similar in concentration to those in the plasma. In a previous study, we measured the concentrations of Na+, K+, and Cl- in condensates and found that these paralleled one another. We suggested that the sum of Na+ and K+ in the plasma could be divided by this sum in the condensate to estimate the dilution of respiratory droplets and solutes by condensed water vapor that accumulated in the condenser. We also attempted to use the conductivity of the condensate to estimate the total concentrations of the respiratory electrolytes in the condensates. This approach was frustrated by the presence of high concentrations of NH4+ in the condensate, much of which represented oral contamination (4, 10, 11). Our objective in this article is to show that conductivity can be used to estimate airway electrolyte concentrations and the dilution of respiratory droplets by water vapor if most of the NH4+ is first removed by lyophilization. Measurements of conductivity can be performed on small volumes of lyophilized samples without consuming the samples. This technique provides an inexpensive and reliable method of estimating the total concentration of ions in the condensates and the dilution of respiratory droplets by the water vapor. A comparison was between estimates of dilution made from conductivity, the total cations, and the urea concentrations of the condensates.
Subjects Condensates were collected from a group of 18 normal adult subjects with no history of lung disease or recent smoking (average age, 26 ± 6 SD; 7 males and 11 females). Condensates were collected by having subjects exhale through a nonrebreathing valve connected to a 64 x 22-mm ID ventilator tubing (Corr-a-Flex 2; Hudson RCI, Temecula, CA) into an inexpensive 66-cm commercial Pyrex Allihn condenser, which was cooled with circulating ice water. The condensate dripped into polycarbonate collection tubes. The subjects breathed into the condensers for approximately 1 hour, yielding a total of approximately 10 ml of condensate. Two groups of studies were performed. The effects of lyophilization on conductivity and NH4+ and electrolyte concentrations were studied in the first group of 11 samples obtained from 11 of the normal subjects. In the second group of 13 sets of samples from 13 normal subjects, the variability of condensate concentrations was studied in two consecutive 30-minute samples. Cations, urea, and conductivity were measured in these samples after lyophilization. Samples were also collected from five subjects with tracheostomies to determine the oral contribution to NH4+ found in the condensates. Six salivary samples and five serum samples were also collected for comparison with condensate concentrations.
Sample Analysis Dilution of respiratory droplets was estimated by dividing the total cation concentration in serum (150 mM) by the total concentration of cations estimated from the conductivity (in µmol/L of NaCl) in the lyophilized samples (see DISCUSSION for justification).
A one-way analysis of variance with the Student-Newman-Keuls test was used to compare mean concentrations of ions and urea concentrations in the second set of studies (which included 28 samples). Correlation coefficients were also calculated from these data (SigmaStat2; Jandel Scientific, San Rafael, CA). Unless otherwise indicated, means and SEs are indicated in the text and figures. These studies were approved by the Human Research Review Committees, and consent was obtained from each subject before each study.
As indicated in Figure 1 , NH4+ concentrations in the condensate averaged 141 ± 26 (SEM) µmol/L and represented 93 ± 3% of the total cations in the condensate samples before lyophilization in the first group of condensate samples obtained from 11 of the normal individuals (three males and eight females). Lyophilization removed more than 99% of the NH4+ and reduced both the sum of the total cations and the conductivity of the condensates. NH4+ concentrations were significantly lower in patients with tracheostomies (26 ± 13 µmol/L, n = 5). This reflects the fact that most of the NH4+ found in the condensates is delivered to the condensates as NH3 gas from the mouth (see DISCUSSION).
In the second group of 28 lyophilized samples, which were obtained from 15 of the normal subjects (six males and nine females), the total cation concentrations averaged 11.7 µmol/L (Figure 2) . Na+, K+, Ca2+, Mg2+, and urea concentrations averaged 55.3 ± 2.3%, 19.1 ± 1.9%, 6.5 ± 0.5%, 1.8 ± 0.3%, and 3.7 ± 1.2% of the concentrations of total nonvolatile cations. Na+ concentrations were significantly greater than those of the other nonvolatile ions and urea. K+ concentrations significantly exceeded those of Ca2+, Mg2+, and NH4+ in the lyophilized samples (p < 0.05).
Considerable variation was found in both the conductivity and total cation concentrations of the samples when paired condensate samples were collected in two consecutive 30-minute intervals in individual subjects (Figure 3) . However, conductivity was significantly correlated with concentrations of total cations, Na+, K+, Ca2+, Mg2+, and urea (Figure 4) . As expected, the correlation coefficients decreased when concentrations of the individual solutes were low.
The total concentration of cations in the serum collected in five normal subjects averaged 149 ± 1.8 mEq/L, with a coefficient of variation of 0.027. Urea concentrations in the serum averaged 4.4 ± 0.5 mmol/L (coefficient of variation = 0.3). The dilution of the respiratory droplets was calculated with Equations 1–3, with the assumption that total plasma concentrations were equal to those measured in serum samples from five normal subjects (serum cation concentration = 150 mM and serum urea concentration = 4.4 mM). As indicated in Figure 5 , the average dilution values calculated with these three indicators of dilution were not significantly different.
The concentrations of each of the cations and urea in the respiratory fluid and droplets were calculated with Equation 1 (Figure 6) . Normal plasma concentrations are indicated in dashed lines. It will be noted that Na+ concentrations were less than those in plasma. K+ and Ca2+ were greater than those in plasma, and Mg2+ and urea are about the same as plasma levels.
Salivary concentrations of NH4+ (4,114 ± 1,209 µmol/L) were similar to those of Na+ (3,876 ± 656 µmol/L) but were significantly less than those of K+ (14,337 ± 1,470 µmol/L) in six subjects (Figure 6). Ca2+ concentrations averaged 1,221 ± 152 µmol/L, and those of Mg2+ averaged 114 ± 14 µmol/L. As noted in our previous study, solute concentrations in the condensates are occasionally very high. This tends to skew mean dilution values to lower levels. Very high concentrations of sodium and conductivity were also found in two samples obtained from a single subject in this series of studies. To avoid a disproportionate increase in the mean data, samples obtained from this one subject were excluded from the reported data. Elimination of these high concentrations reduced average values of ion concentrations below those reported in our previous study (4).
The movement of inflammatory mediators and other solutes from the lung to the airspaces and airways must occur by diffusion and/or convection. Nonvolatile indicators, such as electrolytes, proteins, and polypeptides, can only be transferred into the air phase of the lungs by convective processes, primarily droplet formation. Before the introduction of the exhaled breath condensate procedure, it was generally assumed that all of the water in the exhaled air was generated by evaporation from the respiratory surfaces and that it contained no nonvolatile solutes. However, the presence of trace concentrations of nonvolatile solutes must indicate that there is a small though variable fraction of the exhaled fluid that is generated by droplet formation in the lung, rather than by the formation of water vapor from evaporation from the respiratory surfaces. Concentrations of all of the ions in the condensates were very low, and the sensitivity of measurements was significantly improved by using ion chromatography instead of ion specific electrodes and/or chemical methods used in our previous report (4). NH4+ represented 93% of the cations detected in the condensate of the normal subjects. High concentrations of NH4+ in the condensates reflect both oral production of NH3 and the remarkable solubility of NH3 in water in fluid lining the condenser walls (approximately 50,000:1 at pH 7.4 and 5°C). As in our earlier study and that of Vass and colleagues (10), concentrations were much lower in five subjects with tracheostomies, indicating that most of this volatile cation is derived from the mouth. Rinsing the mouth with acidic fluid (pH 2.5) immediately reduces exhaled NH4+ concentrations by 90% (11). Concentrations of NH4+ collected by various condensers may be influenced by a variety of factors, including local pH, temperature, and arrangement of tubing and whether the condensate is allowed to drip from the condenser during the collection period (4). Lyophilization can successfully remove NH3 from solutions of NH4OH and NH4HCO3 but is much less efficient in removing NH3 from solutions of NH4Cl. More than 99% of the NH4+ is removed from these samples by lyophilization. This observation is consistent with the proposition that most of the NH4+ in the condensate samples was delivered to the condensers as NH3 gas (4). Na+ concentrations in condensate samples varied from less than 1 µmol/L to greater than 30 µmol/L. Considerable variability in Na+ concentrations was also observed in our previous study. Approximately 90% of samples have Na+ concentrations below 20 µmol/L, but occasional individuals have significantly higher Na+ concentrations, which tend to increase mean values and are responsible for the observation that median values are below mean values (4). Significant variation was observed in the previous study, even when samples were collected from the same subjects (4). Furthermore, significant variability was found in both the total cation concentrations and conductivity of condensates that were collected in consecutive 30-minute intervals (Figure 3). Although comparable variability has been observed in the concentrations of protein in the condensate (12) and thromboxane concentrations were also quite variable in patients who collected samples with and without nose clips (10), adenosine concentrations were relatively constant in the study of Vass and colleagues (10). Variability was also modest in day-to-day measurements of a variety of aldehydes, but it is not clear whether these solutes are volatile and might have been delivered in the gaseous phase rather than in the respiratory droplets (6). Despite wide variations in cation concentrations observed in normal subjects, the concentrations of multiple hydrophilic solutes were significantly correlated (Figure 4). Values of the correlation coefficient decreased at lower concentrations, presumably reflecting the decreased reproducibility of the ionic and chemical analyses at low concentrations. These observations are consistent with the conclusion that much of the variation observed in condensate concentrations that was observed in this study was related to variation in formation of droplets from the fluids lining the respiratory surfaces
Because all of the nonvolatile, hydrophilic solutes in the respiratory droplets are diluted by the same factor by condensed water vapor, it should be possible to select appropriate reference indicators of dilution for these molecules. To calculate the dilution of respiratory droplets by water vapor, the concentration of the reference indicator in the respiratory fluid and droplets must be known. Ideally, the concentrations of the reference indicator in the respiratory fluid should equal those in the plasma. In our former publication (4), we suggested that the sum of Na+ and K+ might be used as an estimate of the total cation concentration in the plasma and condensate to calculate dilution (D):
This equation was based on the premise that the respiratory fluid is isotonic and that Na+ and K+ are the principal cations present in the condensate. The inclusion of K+ in this calculation was based on the observation that K+ concentrations are disproportionately high relevant to Na+ when compared with plasma. Application of ion chromatography in this study made it possible to add Ca2+ and Mg2+ to the measured cations. With this additional information, dilution can be calculated from the total measured cations, where [total cations] = [Na+] + [K+] + 2[Ca2+] + 2[Mg2+]. Note that NH4+ concentration is not incorporated in this equation because most of the NH4+ recovered in the condensate arrives as gaseous NH3 generated in the mouth rather than the lungs and therefore provides no information regarding the dilution of the respiratory droplets. Concentrations of Ca2+ and Mg2+ were multiplied by two so that the data could be directly compared with conductivity data. Total cation concentrations remained relatively constant in serum samples (149 ± 1.8 mM, coefficient of variation = 2.7%). It was therefore assumed that the cation concentration in the serum is 150 mM in Equation 2. Equations 1 and 2 are formally based on the assumption that the airway surface fluid is isotonic; in other words, water rapidly equilibrates between the plasma and fluid covering the airway surfaces. There is some evidence that the airway fluid is not isotonic in all subjects, but this remains controversial (13–27). Alterations in the osmolality of the airway fluid would presumably be associated with proportionate changes in the concentrations of inflammatory mediators, and these changes would not be detected from the dilution of the cations or conductivity of the condensate. Although an increase in the concentration of a mediator caused by dehydration of the airway fluid would not reflect an increase in local production, it might have a physiologic effect on the airways.
The reliability of estimates of dilution based on electrolyte concentrations could be checked if comparable data were obtained with an alternative, nonionic indicator that was present in equal concentrations in the plasma and airway fluid. Urea has a number of properties that suggest that it might be an appropriate, nonionic indicator of dilution. There is no evidence for significant production or metabolism of urea in the lungs, and we could find none in a separate study of perfused rat lungs (unpublished). It is not volatile: None is lost during lyophilization. Furthermore, urea readily diffuses between the plasma and airway fluid (28). This property of urea has complicated its use in studies of bronchoalveolar lavage because plasma urea tends to diffuse into the fluid filling the airspaces during lavage, thereby compromising estimates of solute concentrations that were in the airway fluid before lavage (3). This is not a problem in condensate studies because no fluid is introduced into the lungs and the dilution of respiratory droplets can be calculated with this equation:
Measurement of urea concentrations was based on degradation of urea to NH4+ by urease. However, it was first necessary to remove the high concentrations of NH4+ present in the samples by lyophilization before the samples were treated with urease. The average dilution calculated from the ratio of serum urea to condensate urea resembled the dilution of the total cations. This suggests that the respiratory fluids are relatively isotonic in these normal subjects at rest and that the total cations can be used to estimate the dilution of respiratory droplets in condensates in these subjects. Urea concentrations could also be used to estimate dilution; however, concentrations of urea in the condensate are comparatively low, and serum values are more variable than those of the total electrolytes. It is also possible that there may be bacterial degradation of urea in infected lungs. Estimates of dilution with urea are consequently less accurate than those made with electrolytes. Dwyer (29) had to collect 50 ml of condensate over a period of 10 hours from 10 normal adults to measure urea in the condensate samples. These samples were then lyophilized, increasing the concentrations of urea by 250 times, and NH4+ was removed with an ion exchange column. Condensate urea concentrations averaged 0.39 ± 0.08 µmol/L. This is very similar to the value found in this study (0.33 ± 0.04 µmol/L), in which 5-ml samples were sufficient to make these measurements. Further improvements in analysis of urea may make it possible to make these measurements in smaller volumes of fluid and could provide a more precise criterion for determining whether the respiratory fluid has become hypertonic or hypotonic during exercise or various pulmonary disorders. Removal of NH4+ from the condensates also made it possible to use conductivity to estimate dilution of the respiratory droplets by water vapor. Accurate temperature-adjusted conductivity measurements can be made on 1.0 ml of sample, which can then be used again in other assays. The equipment is inexpensive, and measurements can be completed in less than 1 minute. The condensate samples must be first lyophilized, but this has an added advantage: The samples can be concentrated by reconstituting them with reduced volumes of deionized water, improving the sensitivity of the assays. Lyophilization may also help maintain the integrity of some labile mediators. Thus, the dilution of respiratory droplets may be calculated from the total cation concentration of the plasma or serum and the total cation concentration of the condensates estimated from the conductivity of the lyophilized condensate samples with equation (1). Agreement of the conductivity and total cation data indicate that most of the ions in the lyophilized condensate are strong ions (30) and is consistent with our previous observations (4). The sensitivity of the measurement is increased by intentional reconstitution of the lyophilized samples in smaller volumes. Because total plasma and serum concentrations of electrolytes are very constant in normal subjects, collection of blood may not be necessary in subjects who have not had significant electrolyte abnormalities in the recent past. Because lyophilization is readily available in most institutions and conductivity of water is routinely measured with simple and inexpensive meters in most laboratories, this approach can be used as a relatively simple method of correcting condensate data for dilution and determining whether differences in mediator concentrations between population groups are related to droplet dilution or true changes in respiratory fluid concentrations. Dilution determined with each of the indicators averaged approximately 20,000:1. On the basis of his urea determinations, Dwyer (29) estimated that dilution was greater than 10,000:1. A dilution of 20,000:1 is equivalent to the presence of only 0.05 µl of respiratory fluid in each milliliter of condensate. Furthermore, dilution varied from approximately 50,000:1 to 1000:1, indicating that normal subjects can exhale anywhere from 0.02 to 1.0 µl of respiratory fluid in each milliliter of condensate. Both this and our previous study indicate that the relative concentrations of ions in the condensate differ significantly from those in the plasma and saliva, indicating that the solutes captured in the condenser are not directly derived from these sources. Values calculated for Na+ in the airway surface fluid were less than those in plasma, whereas those of K+ and Ca2+ were greater than plasma values. Potassium concentrations exceeded those of any other ion and NH4+ concentrations were greater than those found in plasma or condensate. The limitations of nonvolatile dilutional indicators for interpreting mediator concentrations in condensates must be kept in mind. They cannot provide information regarding the dilution of indicators that are delivered in the gas phase, and should not be used to estimate respiratory concentrations of volatile solutes such as NH3. However, aside from information regarding dilution of nonvolatile solutes, they may provide new insight into respiratory droplet formation in patients with various pulmonary disorders.
The authors thank Stacy Armstrong, Jake Berger, and Amy Adashek for assistance in these studies.
Supported by National Institutes of Health Grants HL60057 and DC03191. Conflict of Interest Statement: R.M.E. has no declared conflict of interest; J.B. has no declared conflict of interest; B.F. has no declared conflict of interest; K.H. has no declared conflict of interest; M.B.D. has no declared conflict of interest; D.C. has no declared conflict of interest; M.B. has no declared conflict of interest; F.S. has no declared conflict of interest; R.S. has no declared conflict of interest. Received in original form July 8, 2003; accepted in final form September 22, 2003
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
METHODS
