INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Keywords: exhaled condensates; respiratory secretions; aerosol droplets; saliva; bronchoalveolar lavage

Bronchoalveolar lavage (BAL) has improved understanding of the pathogenesis of inflammatory lung diseases and has been helpful in detecting and distinguishing these disorders. Performance of BAL is associated with some risk, including impairment of gas exchange and pneumonitis, and it is difficult to repeat BAL at frequent intervals. It would be preferable if the presence of inflammation in the lungs could be noninvasively followed in exhaled air. More than 200 volatile substances have been detected in exhaled air (1), including NO, which has been used to assess pulmonary inflammation in asthma (2). Unfortunately, most inflammatory mediators and other solutes of interest are not volatile and are consequently not present in the gaseous phase of exhaled air. Nevertheless, low concentrations of these solutes have been found in exhaled water collected with cooled condensers. These solutes are presumably present in droplets released from the fluid lining the respiratory membranes. Increased concentrations of inflammatory markers have been reported in the exhaled condensate collected from patients with a variety of pulmonary disorders (Table 1) (3).

We investigated the electrolyte and buffer concentrations of solutes present in the exhaled condensate of normal subjects. The concentrations of these solutes were much lower than those in plasma, ranging from less than 0.01% to as much as 2% of those in plasma. Marked variations in solute concentrations were found among normal subjects and in individual subjects between tests. Similar variability in solute recovery was reported by Scheideler and coworkers (24), who were unable to detect any protein in the condensates of two normal subjects and found protein concentrations between 0.76 µg/ml and 107.7 µg/ml in condensate from eight other healthy individuals. We also found that condensate concentrations of Na⁺, K⁺, and Cl were well correlated with one another over a 200-fold range of concentrations. This observation suggests that dilution of droplets generated in the respiratory tree by water vapor varies considerably. It is consequently impossible to determine whether increases in the concentrations of mediators reported in various disorders reflect increased droplet formation or increases in the concentrations of these mediators in fluid lining the respiratory tract.

All of the nonvolatile solutes in respiratory droplets should be diluted to the same degree by water vapor deposited on the walls of the condenser. We minimized the effect of dilution by dividing condensate solute concentrations by the sum of the concentrations of the principal nonvolatile cations ([Na⁺] + [K⁺]) in the condensate. Furthermore, by assuming that respiratory secretions are isosmotic with plasma, we were able to estimate the actual dilution of the respiratory solutes in the condensate and the concentrations of the solutes in the respiratory fluids (RFs).

In the course of these studies, we found that there are significant amounts of NH₄⁺ in exhaled condensate. Most of this appears to be generated in the mouth, and the efficiency with which NH₄⁺ is collected in the condenser is influenced by such factors as the configuration of the collection tubing and the partial pressure of CO₂ (PCO₂) in the air traversing the condenser. In addition, we observed significant concentrations of lactate, which seemed to be derived from the lungs, in condensate. These two ions appear to have been the principal buffers that were collected, and they may influence the pH of two exhaled condensate. This may be clinically relevant, since it has been reported that the pH of condensate is acidic in patients with exacerbations of asthma (3).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

Twenty subjects (14 males and six females), without histories of pulmonary disease or recent smoking, were studied. The average age was 31.6 yr (range: 22 to 53 yr). Condensates were also collected from three subjects with chronic tracheostomies for obstructive sleep apnea.

Procedures

Subjects exhaled into a glass condenser cooled with circulating ice water for approximately 1 h, until about 10 ml of condensate were collected. Approximately 2 ml of saliva were also collected. Samples were stored in Falcon Blue Max polypropylene centrifuge tubes (Becton-Dickinson, Franklin Lakes, NJ) at 20° C.

We determined the effect of the length of the collector tubing on the collection of NH₄⁺ and Na⁺ by comparing condensates from a single normal subject: five condensate samples were collected with the standard tubing (64 cm long × 22 mm I.D.; Corr-a-Flex 2, Hudson RCI, Temecula, CA) and five samples were collected when the tubing had been reduced to 16 cm in length.

An additional study was conducted to determine the effect of CO₂ on the efficiency of NH₄⁺ collection by the condenser. A dog ventilator was connected to a clinical humidifier that was filled with an isotonic saline solution containing 2 mM NH₄Cl. Room air was pumped at 5 L/min (250 ml per stroke × 20 strokes/min) over the fluid in the humidifier without bubbling. The fluid was kept at a sufficiently high temperature (84° C) to maintain the temperature entering the condenser tubing at 37° C and the humidity at 100%. In six experiments, the air pumped into the condenser contained 6% CO₂, and in six other studies it contained no CO₂. The humidifier solution was buffered at pH 7.4 with 25 mM NaHCO₃ in the CO₂ experiments, and with 12 mM 4-(2-hydroxyethyl)-1-piperazine-N-2-ethanesulfonic acid in the air experiments.

Chemical Analyses

Amylase was measured with a modification of a kit (577) from Sigma (St. Louis, MO), lactate oxidase with another kit (735-10) from Sigma, and lactic dehydrogenase with a third Sigma kit (826B). The Protein Micro BCA protein assay reagent kit (Pierce, Rockford, IL) was used to measure protein with the bicinchoninic acid method. Nitrate and nitrite were quantitated through a modified Griess reaction. Coefficients of variation (CV) of conductivity, electrolytes, and NH₄⁺ are shown in Table 2.

Calculations

The dilution (D) of respiratory droplets by water vapor in the condensate was calculated with the equation:

(1)

In this group of normal subjects, it was assumed that [Na⁺]_plasma + [K⁺]_plasma = 144 mM. Concentrations of solutes ([X]_RF) in RF were calculated from the product of their concentration in the condensate and the dilution:

(2)

Statistical Analysis

Friedman's repeated-measures analysis of variance (ANOVA) on ranks, with the Student-Newman-Keul's method of comparing pairs of values, were used to analyze the condensate data (SigmaStat 2; Jandel Scientific, San Rafael, CA), and a value of p < 0.05 was considered significant. One-way ANOVA with Tukey's test was used to analyze saliva data. Linear regression and correlation coefficients were calculated with the same program.

Studies of human subjects were approved by the Human Research Review Committee of the Zablocki VA Hospital and the Medical College of Wisconsin.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Electrolytes

As indicated in Figure 1, the average conductivity of the condensate was equivalent to 497 ± 34 (mean ± SEM) µM NaCl. Ammonium represented a significant fraction of the cations collected ([NH₄⁺]_cond = 229 ± 49 µM, Na⁺ = 242 ± 140 µM, and K⁺ = 80 ± 36 µM). Chloride concentrations averaged 231 ± 109 µM, and lactate averaged 393 ± 305 µM. (Only six samples were analyzed for lactate and eight samples were analyzed for protein. To avoid misinterpretation related to differences in sample sizes, we chose not to include the lactate and protein data in the histogram shown in Figure 1.) Values of conductivity were significantly greater (p < 0.05, as judged with a repeated measures, nonparametric test) than those of NH₄⁺, Na⁺, K⁺, and Cl. Concentrations of NH₄⁺ were significantly greater than those of K⁺, Cl, and Na⁺ by the same criteria. Differences between mean values and median values (indicated by filled circles in Figure 1) were due to variability in the data.

View larger version (31K): [in this window] [in a new window]   Figure 1.   Conductivity and ion concentrations in exhaled condensate. Mean and SEM are indicated. Median values are shown as closed circles. Discrepancies between mean and median values reflect the observation that high values of solutes are sometimes collected in the condensate. Arrows above the bars show were signficant differences exist between ranked values (p < 0.05). Much of the measured conductivity could be accounted for by the high concentrations of NH4+ in the collected condensate.

Variability was seen among subjects, but it was also observed in samples from four subjects collected at different times over the course of several days (Figure 2). We were unable to find any cause for this variation, but good correlations were found between concentrations of Na⁺, Cl, and K⁺ (Figure 3, upper two panels). The variability of the data was reduced when the concentrations of solutes were divided by the sum of the Na⁺ and K⁺ concentrations (note that the mean and median values are similar and that standard errors are relatively small in the first four bars in the histogram shown in Figure 4). NH₄⁺ concentrations in the condensate were poorly coordinated with those of Na⁺ (Figure 3, bottom panel ).

View larger version (14K): [in this window] [in a new window]   Figure 2.   Variability in Na+ concentrations in exhaled condensates collected at different times over several days from individual subjects.

View larger version (18K): [in this window] [in a new window]   Figure 3.   Correlations between ion concentrations. Good correlation was observed between Na+, K+, and Cl, but no significant correlation was observed between Na+ and NH4+.

View larger version (40K): [in this window] [in a new window]   Figure 4.   Ionic ratios. Concentrations of Na+, K+, Cl, and lactate have been divided by the sum of Na+ and K+ concentrations in the first four bars. Note that median (closed circles) and mean values are relatively similar. The sum of the Cl and lactate concentrations is approximately the same as that of the Na+ and K+ concentrations (fifth bar), suggesting that Cl and lactate are the principal anions in condensate. Most of the conductivity could be accounted for by the sum of the measured cations (and associated anions, sixth bar).

Total concentrations of the measured nonvolatile cations were approximately the same as those of anions: the sum of the lactate and Cl concentrations equaled that of the Na⁺ and K⁺ concentrations (Figure 4, fifth bar of the histogram). Furthermore, most of the conductivity could be accounted for by the measured cations (and their associated anions): the sum of the NH₄⁺, Na⁺, and K⁺ (with associated anions) concentrations accounted for virtually all of the observed conductivity (Figure 4, sixth bar of the histogram). If it is assumed that the normal sum of the plasma Na⁺ and K⁺ concentrations is 144 mM, then the dilution of RF by the water vapor collected in the condenser averaged 1:2,443 ± 453.

Using Equation 2, we calculated values for the concentrations of ions and protein in RF. These were [Na⁺]_RF = 91 ± 8 mM, [K⁺]_RF = 60 ± 11 mM, [Cl]_RF = 102 ± 17 mM, [lactate]_RF = 44 ± 17 mM, and [protein]_RF = 7.63 ± 1.82 g/dl.

Concentrations of NH₄⁺ in the condensate averaged 229 ± 43 µM for normal subjects, but no NH₄⁺ was detectable in the condensates collected from tracheostomies in three subjects with obstructive sleep apnea. Na⁺ concentrations in the condensates of these individuals ranged from 25 to 1,385 µM, and lactate concentrations ranged from 0 to 85 µM. Recovery of NH₄⁺ in the exhaled condensate was enhanced by decreasing the dead space between the mouthpiece and the condenser by 75% (Figure 5). This did not have any effect on the recovery of Na⁺. Incoproration of 6% CO₂ in air pumped through the condenser also increased the recovery of NH₄⁺ (Figure 6).

View larger version (21K): [in this window] [in a new window]   Figure 5.   Effect of reducing tubing length on conductivity and NH4+ and Na+ concentrations in condensate. Decreasing the dead space volume increased the efficiency with which NH4+ was recovered in the condensate. This was presumably responsible for the similar effect on conductivity. Note that there was no effect on the recovery of Na+. Reduced recovery of NH4+ in condensate with long tubing is attributable to trapping of exhaled NH3 in small volumes of water lining the tubing.  Short; long.

View larger version (19K): [in this window] [in a new window]   Figure 6.   Effect of passing 6% CO2 during the collection of NH3. Air or air and 5% CO2 were pumped over a heated reservoir that contained a solution of NH4Cl in buffered saline and then through the cooled condenser with a mechanical respirator. Concentrations in NH4+ in the heated reservoir and condensate are indicated. Addition of CO2 to the stream of air significantly improved the efficiency with which NH4+ was trapped by the condenser. Concentrations of NH4+ in the reservoir remained unchanged.  Reservoir;  condensate.

Concentrations of both NO₂ and NO₃ were comparatively very small as compared with those of the other measured ions (NO₂ = 0.76 ± 0.14 µM, n = 14; NO₃ = 4.3 ± 1.0 µM, n = 15). Amylase was undetectable in the condensates of five of the subjects and was present in very low concentrations in the condensates of the others; with a mean concentration of 4.67 units/ml, or less than 2 × 10⁵ that in the saliva. Protein concentrations averaged 2.3 mg/dl, or 1.2 × 10³ of the concentrations in the saliva.

As indicated in Figure 7, salivary solute concentrations were much greater than condensate concentrations for all measured solutes (note that concentrations of the electrolytes are indicated in units of mM rather than µM). Concentrations of K⁺ and Cl were significantly greater than those of Na⁺, NH₄⁺, and lactate.

View larger version (31K): [in this window] [in a new window]   Figure 7.   Salivary solute concentrations. Significant differences (p < 0.05) are indicated by arrows above the bars. Note that salivary solute concentrations are indicated in millimoles/liter, whereas condensate concentrations (Figures 1, 3, 5, and 6) were indicated in micromoles/liter. Relative concentrations of solutes in saliva differed from those found in condensate. Concentrations of NH4+ and lactate in saliva were very low as compared with those of electrolytes. Potassium concentrations exceeded those of Na+. Amylase concentrations were very high as compared with those found in condensate (values are indicated as multiples of 10,000).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The saturation of exhaled air with water normally exceeds 90%, and the rate of water loss by this route is largely governed by the rate of ventilation. Most of the water lost from the lungs is formed as water vapor, a gas that does not contain nonvolatile solutes. As indicated in Figure 8, the only way that nonvolatile solutes can be lost from the surface of the lungs is in the form of respiratory droplets. Droplets formed in this fashion can be deposited on the walls of a condenser, but most of the water that accumulates on the condenser surfaces is derived from water vapor rather than from droplets generated from the secretions lining the pulmonary surfaces. Within condensate samples, the dilution by water vapor of the solute concentrations in respiratory droplets can be estimated from the concentrations of nonvolatile constituents in the condensate.

View larger version (32K): [in this window] [in a new window]   Figure 8.   Respiratory droplets are released from the surfaces of the airways/air spaces. Much greater quantities are released in the form of water vapor (dotted arrows). When the respiratory droplets reach the condenser, they become diluted by large volumes of water vapor that become deposited on the walls of the condenser. Concentrated; dilute.

Nonvolatile Solutes

Despite considerable variation in normal electrolyte concentrations in condensates of respiratory exhalations, we found good correlations between condensate concentrations of Na⁺, K⁺, and Cl. This observation is consistent with the proposition that dilution of respiratory droplets by water vapor deposited in the condensate results in equivalent dilution of each of these electrolytes. Our calculation of the dilution of respiratory droplets was based on two assumptions. The first of these was that respiratory secretions have the same osmolality as plasma. This premise is supported by the observation by Jayaraman and coworkers (25) that the fluids lining cultures of tracheal epithelial cells, live mouse tracheas, and freshly harvested human bronchi are isosmotic. The second assumption in our calculation was that the sum of the Na⁺ and K⁺ concentrations in the RFs is equal to that in the plasma. This postulate is justified by these electrolytes and associated anions being the principal solutes in plasma and cellular fluid, and by the observation that they appear to represent the principal nonvolatile, respiratory cations in respiratory condensate. The dilution of RF by condensate averaged 1:2443. It can therefore be concluded that of the 500 ml of water exhaled in a 24-h period, only 200 µl of RF is lost in the form of droplets each day by normal subjects.

On the assumption that the sum of the plasma Na⁺ and K⁺ concentrations is 144 mM, the following values were calculated from Equation 2 for these ions in the RF: [Na⁺]_RF = 91 ± 8 (mean ± SEM) mM, [K⁺]_RF = 60 ± 11 mM, and [Cl]_RF = 102 ± 17 mM. In the study by Jayaraman and colleagues (25), human bronchi were covered with a layer of fluid which was 55 ± 5 µm thick and contained [Na⁺] = 103 ± 3 mM and [Cl] = 92 ± 4 mM. This would represent a somewhat hypotonic fluid, but Jayaraman and colleagues were unable to measure [K⁺], which averaged about 60% of [Na⁺] in the present study of exhaled condensates. K⁺ concentrations in plasma are much lower than this (approximately 4% of Na⁺ concentrations), but are greater than in plasma in both BAL fluid collected from animals (26) and in the fluid secreted by fetal animals (27).

Lactate concentrations in the condensate were 32% of the sum of the Na⁺ and K⁺ concentrations. This is considerably greater than in plasma, in which lactate concentrations are less than 1% of the sum of the Na⁺ and K⁺ concentrations. It has been noted that much of the glucose metabolized by the lungs is converted to lactate rather than to CO₂ despite high ambient oxygen concentrations in lung tissues (28). This may reflect a relative deficiency of mitochondria in regions with attenuated numbers of pulmonary alveolar type I cells. Concentrations of K⁺ in saliva were significantly higher than those of Na⁺ in the study just cited (28), as well as in other studies (29, 30).

The conductivity of the condensate was expressed in terms of micromoles per liter of NaCl. The sum of [Na⁺], [K⁺], and [NH₄⁺] (and their associated anions) accounted for virtually all of the conductivity in these studies. The sum of the Cl and lactate concentrations was approximately the same as that of the Na⁺ and K⁺ concentrations (Figure 4, fourth bar in the histogram). These four ions appear to be the principal stable ions in respiratory condensate. Smaller concentrations of nitrate and nitrite were also found. These were comparable to those reported by Hunt and associates (21).

Concentrations of amylase in the condensates from normal subjects were frequently undetectable and were disproportionately lower than those in saliva. The average amylase concentration in the condensate was less than 0.01% of that in saliva. In contrast, the condensate Na⁺ concentration averaged 32% of that in saliva. These observations indicate that relatively few of the droplets captured in condensate were formed in the mouth and most presumably came from elsewhere in the respiratory tract. Filtration of exhaled air (e.g., as in the study by Hunt and colleagues [3]) is therefore not needed to remove oral fluids from the condensates of normal subjects. Filtration might actually decrease the efficiency of collection of respiratory secretions in condensate samples, and there is no evidence that a filter with a specific pore size would selectively remove droplets produced in the mouth.

The calculated values of protein in RF were variable but quite high (7.29 ± 4.29 g/dl), approximating those in plasma.

Ammonium

Concentrations of NH₄⁺ (229 ± 43 µM) in condensate exceeded those of Na⁺ and K⁺ and were significantly greater than that found in blood (normally between 6 µM and 47 µM) (31). Since no NH₄⁺ was detectable in the exhaled condensates collected from patients with tracheostomies, it appears that most of the NH₄⁺ in condensate is actually formed in the mouth. Oral NH₄⁺ is generated in part by bacterial hydrolysis of urea in the oral cavity (32), but secretion of NH₄⁺ by salivary and respiratory glands is possible. Normal concentrations of NH₃ in exhaled air are very variable, ranging between 0.1 ppb and 100 ppb, and are not closely linked to blood ammonia levels. For example, breath NH₃ levels in uremic subjects were from five to 20 times greater than normal, even though blood ammonia levels were well within normal range (38). Much of the exhaled NH₃ is actually formed in the mouth. Concentrations of NH₄⁺ in saliva averaged 1,189 µM in the present study and 2,500 µM in another report (41).

We found three reasons for concluding that most of the NH₄⁺ trapped within the condenser in our study was delivered there as NH₃ gas rather than as NH₄⁺ dissolved in droplets. The first reason for this was the observation that increasing the P_CO2 of air from 0 to 6% increased the concentrations of NH₄⁺ captured in the condensate (Figure 6). This can be explained by more efficient trapping of NH₃ gas as NH₄⁺ by acidic droplets (Figure 9) (42). For example, if the concentration of NH₃ is the same on the respiratory or oral membranes, m, where it is generated, and in the condenser, then

(3)

(4)

View larger version (13K): [in this window] [in a new window]   Figure 9.   Enhanced trapping of NH4+ by CO2 in the condenser. Conversion of NH3 to NH4+ in the fluid lining the condenser is increased when this fluid is acidified by CO2 flowing through the condenser (see Figure 6).

where K_a designates the acid dissociation constant of NH₄⁺ (K_a = 9.25).

Thus, if the pH of the droplets in the condenser is 1 unit less than in the fluids lining the respiratory membranes, then once an equilibrium is reached, the concentrations of NH₄⁺ in the condenser droplets should become 10 times greater than those on the respiratory membranes, where the NH₄⁺ was formed. Concentrations of CO₂ in the condenser and droplets at the end of exhalation should reach about 1,250 µM (equivalent to 40 mm Hg). Because this is much greater than the concentrations of NH₃, or of any other buffer in the condensate, the pH of the droplets in the condenser is presumably governed by the P_CO2 of the expired air. At a normal P_CO2 of 40 mm Hg, the pH of an unbuffered aqueous solution is below 6 (43), and efficient "trapping" of NH₄⁺ in the condensate would therefore be expected.

The second reason for concluding that NH₄⁺ trapped in the condenser was delivered in the form of NH₃ gas is the effect of tubing length on concentrations of NH₄⁺ found in the condensate (Figure 5). NH₄⁺ concentrations in the condensate were greater when the length of the tubing between the mouthpiece and condenser was reduced. This presumably reflects some trapping in the small amount of fluid that remains on the surfaces of this tubing. Concentrations of Na⁺ in the condensate were not influenced by this change because Na⁺ was delivered proportionately with water droplets throughout the tubing and condenser.

The third reason for proposing that NH₄⁺ collected in the condensate was delivered as a gas rather than in droplets was the observation that, although concentrations of electrolytes were well correlated with one another, correlations between Na⁺ and NH₄⁺ concentrations in the condensate were very poor (Figure 3). This indicates that much of the NH₄⁺ in the condensate was not delivered as droplets from the airways.

If it were assumed that most of the NH₃ in exhaled air is captured in the condensate, then the amount of NH₃ lost each day in the exhalate would amount to about 115 µmol. This represents significantly less than 0.1% of the nitrogen lost as urea in the urine each day (0.5 moles) and less than 0.3% of the amount lost as urine NH₄⁺ (approximately 35 mmol when the urine has a pH of 5.0) (44).

Following collection, the condensate in our studies was exposed to air, and it is likely that most of the CO₂ was lost. Hunt and colleagues (3) degassed their samples with argon, a procedure that should have removed virtually all of the CO₂. However, removal of NH₃ from solutions can only be accomplished if the solutions are initially alkalinized to pH values above 10, and concentrations of NH₄⁺ in both the study by Hunt and colleagues and in the present study were presumably relatively constant after collection (45). The concentration of NH₄⁺ trapped in the condensate will be determined in part by the original P_CO2 of the exhaled air. It is therefore possible that the low values of pH found in asthmatic subjects by Hunt and colleagues partly reflect the increased dead space and hyperventilation in asthmatic subjects. These abnormalities of ventilation would decrease the end-tidal P_CO2 and the P_CO2 in the condensate. This in turn would reduce the amount of NH₄⁺ that was trapped. Once CO₂ was purged from the asthmatic samples with argon, the pH of the condensate would be relatively acid because it contained less NH₄⁺. However, it is also possible that the production of NH₃ is reduced in asthma or that an acid species, such as lactic acid, is secreted into the airway fluid during asthmatic attacks. Nevertheless, interpretation of condensate pH is complicated by the partial dependency of NH₄⁺ concentrations in the condensate on conditions in the mouth and by the efficiency with which NH₃ is trapped in the condenser. It is also complicated by the extremely low concentrations of acids, bases, and buffers found in exhaled water (43, 46).

Concentrations and Concentration Ratios

Much of the variation found in nonvolatile solute concentrations in expiratory condensates is probably related to differences in the dilution of respiratory droplets by the water of vaporization. This is a serious problem, since increases in concentrations of individual mediators may reflect increased droplet formation rather than increases in mediator concentrations in the RF. Increased droplet formation could be associated with clinical evidence for bubbling in the respiratory tree (e.g., rales, wheezes, and rhonchi). It is possible to minimize differences in solute concentrations related to droplet formation by calculating the ratio of relative concentrations of these solutes to the concentration of some other solute. For example, the ratio of oxidized to reduced glutathione may yield useful data concerning redox potentials in tissue. Interpretation of changes in the concentrations of individual indicators in the condensate cannot be reliably made unless the dilution of the RF by exhaled water vapor is known. This dilution can be estimated from the concentrations of one or more reference indicators that are present in similar concentrations in RF and plasma. For reasons described previously, we have chosen the sum of the Na⁺ and K⁺ concentrations for this purpose. Other indicators, such as urea, might be used instead, since concentrations of urea tend to be similar in most body compartments. Unfortunately, it is difficult to measure small concentrations of urea in exhalate, in part because of the presence of high concentrations of ammonia, which is also the product of enzymatic assays of urea.

Similar problems with dilution have been associated with BAL, since the RF are diluted in the relatively large volume of lavage fluid, which characteristically consists of saline. One of the important advantages of the condensate approach is that the RF on the surface of the lung remains undisturbed and undiluted. This avoids sudden shifts of solutes from the tissues to airspaces that occur during lavage (26). The condensate approach also makes it possible to measure Na⁺ and Cl concentrations, which is not possible when the lungs are lavaged with saline solutions. However, concentrations of all solutes in exhaled condensates are often extremely low, and measurements of the concentrations of electrolytes and other solutes may be very difficult to make with sufficient accuracy. This is illustrated in Table 2, which shows how the reproducibility of such measurements tends to decrease at low concentrations. It should be emphasized that even a CV as great as 15% would represent a considerable improvement over variation related to droplet dilution, which can be more than 100-fold (i.e., 10⁴%). Studies of condensates should include similar information about the reproducibility of measurements of the solutes, including the "reference" indicators.

If the quantity of droplets in exhaled condensate is very low, and concentrations of the reference indicators (e.g., Na⁺ + K⁺) cannot be determined in a reliable manner, then interpretation of the concentrations of any other solutes (e.g., inflammatory mediators) becomes problematic. Regardless of the accuracy with which inflammatory mediators can be measured, changes in their concentrations in condensate may vary by a factor of 100 or more, depending on variations in the dilution of respiratory droplets by the water of vaporization. Methods should be developed to increase respiratory droplet production, so that more reliable estimates of solute dilution can be made. In addition, methods for analyzing very low concentrations of reference and other indicators should be pursued.