Correcting for tissue nitrogen excretion in multiple breath washout measurements

Nitrogen excreted from body tissues impacts the calculation of multiple breath nitrogen washout (MBWN2) outcomes. The aim of this study was to determine the effect of tissue N2 on MBWN2 outcomes in both healthy subjects and patients with CF and to assess whether it is possible to correct for tissue N2. The contribution of tissue N2 to MBWN2 outcomes was estimated by comparing MBWN2-derived functional residual capacity (FRCN2) to FRC measured by body plethysmography (FRCpleth) and by comparing MBW outcome measures derived from MBWN2 and sulfur hexafluoride MBW (MBWSF6). Compared to plethysmography and MBWSF6, MBWN2 overestimated FRC and lung clearance index (LCI). Application of mathematical tissue N2 corrections reduced FRCN2 values closer to FRCpleth in health and reduced LCIN2 in both health and CF, but did not explain all of the differences observed between N2-dependent and -independent techniques. Use of earlier washout cut-offs could reduce the influence of tissue N2. Applying tissue N2 corrections to LCIN2 measurements did not significantly affect the interpretation of treatment effects reported in a previously published interventional trial. While tissue N2 excretion likely has an impact on MBWN2 outcomes, better understanding of the nature of this phenomenon is required before routine correction can be implemented into current MBWN2 protocols.


Introduction
Multiple breath nitrogen washout (MBW N2 ) has been shown to be a feasible and sensitive test to measure ventilation inhomogeneity and detect early obstructive lung disease in children and adults [1,2]. Nitrogen (N 2 ) excreted from body tissues through the lungs can impact the calculation of MBW N2 outcomes, including the functional residual capacity (FRC) and lung clearance index (LCI) [3,4]. Several studies have measured the elimination of tissue N 2 in healthy adults from its accumulation during breathing of 100% oxygen for prolonged periods [5][6][7][8][9][10][11]. Based on these studies, the tissue N 2 excretion rate and accumulated volume over time was found to fit a multi-phase exponential curve with the early phases representing the desaturation of highly perfused tissues and the later phases representing the slower desaturation of poorly-circulated and fat-containing tissues. Elimination rates were found to vary both within and between individuals. Recently, Nielsen et al. applied a tissue N 2 excretion equation to a simulated washout in a two compartment lung model with variable dead space and ventilation heterogeneity [3]. Yammine et al. used a different approach to illustrate the effect of tissue N 2 on the washout by subtracting 1% end-tidal concentration of N 2 evenly over the course of the washout for one healthy subject and one subject with cystic fibrosis (CF) [4]. These two studies confirmed that there is a greater effect of tissue N 2 on MBW N2 outcomes in disease versus health, but they did not explore whether the contribution of tissue N 2 can be adequately offset in measurements from subjects with a range of body size and lung disease severity. In patients with CF, increased ventilation inhomogeneity leads to greater washout duration, and in theory, longer washouts have a greater total contribution of tissue N 2 . Therefore, the impact of tissue N 2 excretion likely introduces greater bias in a subject with significant lung disease compared to a healthy subject of similar size and leads to the overestimation of their FRC and other MBW N2 outcomes [2][3][4]12].
There are limited data to support correcting for the contribution of tissue N 2 ; thus it is not currently recommended as per American Thoracic Society/European Respiratory Society (ATS/ERS) consensus statement [12]. As MBW N2 develops into an increasingly important clinical research tool for the monitoring of CF lung disease and the assessment of treatment effects, the role of tissue N 2 must be clarified in order to determine whether it is necessary to correct for its contribution to the MBW N2 test. The aim of this study was to estimate the magnitude of tissue N 2 in both healthy pediatric and adult subjects and patients with CF across a range of disease severity and to assess the effect of applying correction factors for tissue N 2 on the MBW N2 test and on treatment effects in interventional trials.

Study participants
Data were collected as part of four previously published studies [2,[13][14][15]. Healthy participants without a history of respiratory disease or current acute respiratory tract symptoms were recruited from staff and families at the Hospital for Sick Children. Participants with a confirmed diagnosis of CF (defined by a positive newborn screening test or at least one clinical feature of CF in combination with either a documented sweat chloride >60 mEq/L by quantitative pilocarpine iontophoresis test or a genotype with two CF-causing mutations) were recruited from families attending a routine visit to the CF outpatient clinic at the Hospital for Sick Children or St. Michael's Hospital in Toronto, Canada. Informed written consent was obtained from the participant or parent/guardian for all subjects. The original studies were approved by the Research Ethics Board at the Hospital for Sick Children (REB #1000019945, #1000024909, and #1000023162) and St. Michael's Hospital (REB #12-139), Toronto, Canada.
Pulmonary function testing MBW N2 measurements were performed using an open circuit, bias flow system (Exhalyzer D1, EcoMedics AG, Duernten, Switzerland) and associated software (Spiroware1 3.1 Eco-Medics AG). A subgroup of subjects also performed MBW tests using a respiratory mass spectrometer system (AMIS 2000, Innovision A/S, Odense, Denmark), which used sulfur hexafluoride (SF 6 ) as the tracer gas. MBW SF6 traces were analyzed by a single trained observer using custom-written analysis software (TestPoint, Capital Equipment Corp., Billerica, MA, USA). All MBW trials were reviewed for quality control according to guidelines proposed in the ATS/ERS consensus statement [12]. In addition to MBW testing, subjects performed plethysmographic lung volume measurements using the Vmax system (VIASYS CareFusion, San Diego, California, USA) according to ATS standards [16].

Estimates of tissue N 2 contribution
FRC measured by body plethysmograph (FRC pleth ) includes the volume of all compressible intrathoracic gas, whereas only the volume of communicating lung units is measured during MBW. Therefore, in healthy individuals, FRC measured by a gas-dilution technique (such as MBW N2 ) should be equal to or less than that measured by plethysmography [17] in the absence of endogenous production of the tracer gas. Thus the differences between FRC pleth and FRC N2 can be used to approximate the contribution of tissue N 2 to the MBW N2 . Similarly, as SF 6 is an exogenous, biologically inert gas that does not dissolve significantly in blood or other tissues, it was used as an indirect reference method to assess the magnitude of the contribution of tissue N 2 to FRC derived by gas dilution.
Tissue N 2 excretion equations MBW N2 assesses ventilation inhomogeneity by examining N 2 clearance over a series of breaths for the duration of the washout. To generate MBW N2 outcomes, the total volume of exhaled gas (net cumulative expired volume; CEV) and the total volume of inert gas expired per breath (cumulative expired volume of N 2 ; CEV N2 ) must be measured. FRC and LCI are calculated when Cet N2 falls below a predefined threshold (typically 2.5% of the initial CetN2).
where Cet N2 is the end tidal concentration of nitrogen. Cet N2, initial is the end tidal concentration of N 2 in the first breath of the washout phase, and Cet N2, final is the end tidal concentration of N 2 in the first breath of the washout phase where Cet N2 is less than the target threshold. DS pre is the equipment deadspace proximal to the sampling point of the apparatus. In order to correct these values for tissue N 2 excretion, breath-by-breath end tidal body tissue N 2 concentration (Cet N2 BT ) as well as the volume of body tissue nitrogen excreted over the washout (V N2 BT ) are subtracted from Eqs 1 and 2 (Eqs 3-5). The volume of tissue nitrogen was generated for the entire breath (from the start of inhalation to the end of exhalation).
where V N2,BTi is the volume of body tissue nitrogen expired in breath i and VExp i is the net volume of expired gas in breath i. Cet N2 is the end tidal concentration of nitrogen. Cet N2 BT is the end tidal concentration of nitrogen derived from the body tissues. Initial subscript indicates the first breath of the washout phase, and final subscript indicates the first breath of the washout phase where (Cet N2 -Cet N2 BT ) is less than the target threshold. Three different equations ( Table 1, Fig 1) were used to derive (V N2 BT ). Cournand's body size-dependent Eq (6) and Lundin's three-phase exponential excretion rate Eq (14) are timedependent and calculate the end tidal tissue N 2 concentration (Cet N2 BT ). The ATS/ERS (22) equation is time-independent and is therefore only used to generate V N2 BT and not Cet N2 BT . Therefore, corrected FRC values were generated from all three equations (with Cet N2, final being uncorrected in the ATS/ERS equation), but corrected LCI values were only generated from the time-dependent equations.
To assess whether the breath-by-breath calculated FRC achieves a plateau, linear regression slopes of the FRC N2 /time curves were calculated for the second half all uncorrected and corrected washouts.
Comparisons of the corrected and uncorrected FRC and LCI results were made with FRC pleth and the difference in FRC and LCI measured by MBW N2 and MBWSF 6 , when available. FRC and LCI values were also re-calculated from the Cournand and Lundin-corrected measurements at the standard MBW end-point of 2.5% normalized end-tidal N 2 concentration, as well as for earlier end-points of 5%, 9%, 12%, and 18% normalized end-tidal N 2 concentration. These end-points were chosen to reflect previous studies that evaluated earlier cutoffs and existing software algorithms [18].

Accuracy of derived nitrogen concentration
Since N 2 concentration values generated by the Exhalyzer D are derived from O 2 and CO 2 concentrations and not directly measured, our results may be biased if these derived values are inaccurate . To ensure the accuracy of the derived N 2 values over the range observed during a MBW N2 test, we compared the C ET N 2 calculated by the Spiroware software to a set of reference gases generated by blending medical air (compressed on site with presumed gas concentrations: F 1 CO 2 = 0.0004, F 1 O 2 = 0.2095, F 1 N 2 = 0.7808, F 1Ar = 0.0093) with a high precision gas mixture (F 2 CO 2 = 0.0500, F 2 O 2 = 0.9500; Praxair Canada, Mississauga ON). FN 2 of the mixed reference gas (F M N 2 ) was calculated using Dalton's Law of partial pressures, the fractional concentrations of the reference gases and the measured FO 2 of the mixed gas (F M O 2 ) . F M O 2 was measured using the Oxigraf laser oxygen analyzer (Oxigraf Inc, Sunnyvale CA, USA) within the Exhalyzer D 1 . The accuracy of the Oxigraf analyzer was confirmed against a Integrating to derive V N2 at time t:

Statistical analysis
Study population characteristics and lung function measurements were summarized as mean and standard deviation (SD). Group differences were calculated using two-sample t-tests, whereas differences in outcomes within the same subject were compared using paired t-tests. The agreement between outcomes within the same subject was assessed using Bland-Altman plots. Pearson correlations were used to determine the correlation between two outcomes. All statistical analysis was conducted using R version 3.1.2 (R Foundation for Statistical Computing, Vienna, Austria).

Accuracy of derived nitrogen concentration
The absolute difference between FN 2 reported by the Exhalyzer D and the reference concentrations (F M N 2 ) was measured over the full range of washout nitrogen concentrations. The mean absolute difference was 0.064% (95% CI -0.032 to 0.16). All measured differences (n = 14) were less than 0.12%. Therefore, we considered the CetN 2 derived by the Exhalyzer D to accurately reflect the true CetN 2 .

Estimates of tissue N 2 contribution to FRC
Characteristics of study participants included are shown in Table 2. Healthy subjects and individuals with CF did not differ in age or lung volumes measured by either MBW N2 or body plethysmography. As expected, LCI measured by MBW N2 was significantly higher in patients with CF. FRC measured by the MBW N2 gas dilution technique (FRC N2 ) should be smaller than or equal to, but not exceed, FRC measured by body plethysmography (FRC pleth ). However, healthy subjects who performed both techniques had FRC N2 values that were on average greater than FRC pleth (mean difference 0.21L; 95% CI 0.12 to 0.29, p<0.001). In contrast, the relationship between FRC N2 and FRC pleth was inconsistent in subjects with CF (mean difference 0.06; 95% CI -0.10 to 0.21, p = 0.44). FRC N2 values were recalculated by applying the three tissue N 2 excretion equations. Application of all three tissue N 2 excretion equations decreased FRC N2 values compared to FRC pleth in health and CF (Fig 2).
Given that the Cournand and Lundin excretion equations improve the FRC N2 agreement with plethysmography, the uncorrected FRC N2 (FRC uncorr ) and the FRC N2 corrected (FRC Cournand and FRC Lundin ) were then compared within subjects ( Table 3).
The within-subject difference in FRC as measured by MBW N2 and MBW SF6 (FRC N2 -FRC SF6 ) were also compared with the estimated contribution of tissue N 2 to FRC N2 . The difference between FRC N2 and FRC SF6 was positively correlated with increased washout time (r = 0.69, p<0.001). FRC N2 became disproportionately larger than FRC SF6 as the contribution of tissue N 2 as estimated by FRC uncorr −FRC Cournand increased (r = 0.68, p<0.001) (Fig 3).
When plotted against washout time, the breath-by-breath calculation of FRC N2 did not plateau as would be expected in a closed system, but rather continued to increase throughout the washout (representative examples from health and CF shown in Fig 4). This is consistent with continuous tissue N 2 excretion. Breath-by-breath correction of the FRC N2 values using the Cournand and Lundin equations decreased the rate of rise of the FRC N2 by 23-34%, but did  not reduce it to zero (Fig 4, Table 4). The absolute and relative magnitudes of the decrease in the FRC/time slope were greater in healthy subjects than in those with CF for both the Lundin and Cournand equations (Table 4).

Impact of tissue N 2 at earlier washout cut-offs
With application of the Cournand tissue N 2 excretion equation, the effect of tissue N 2 (LCI uncorr −LCI Cournand ) decreased when LCI N2 was calculated at earlier cut-offs of the washout (Fig 5). Compared to the traditional cut-off of 2.5% normalized end-tidal concentration of N 2 , the difference between corrected and uncorrected LCI (LCI uncorr −LCI Cournand ) was less pronounced at the 5% cut-off and was no longer significant by the 9% cut-off. While the effect of tissue N 2 (LCI uncorr −LCI Cournand ) on LCI N2 calculated at the 2.5% cut-off increased as disease severity (LCI N2 ) increased (r = 0.61, p<0.001) (Fig 6A), this relationship was not observed at the 5% cut-off (r = 0.17, p = 0.13) (Fig 6B).

Impact of tissue N 2 correction on interventional trial outcomes
Both the Cournand and Lundin equations were applied to MBW data of an observational study investigating the effect of ivacaftor on LCI in children with class 3 mutations in CF [14] ( Table 5). The Lundin-corrected treatment effect was significantly smaller than the uncorrected value (p = 0.01) and the Cournand-corrected difference showed a similar trend  Table 4.
https://doi.org/10.1371/journal.pone.0185553.g004 Fig 4) for healthy subjects and those with CF. Average paired difference (uncorrected-corrected) in absolute and relative (percent of uncorrected slope) terms are shown. Data are expressed as mean 95% confidence interval) unless otherwise stated.

Table 4. Average slopes of the second half of all uncorrected, Lundin-corrected and Cournand-corrected FRC N2 /breath number curves (depicted graphically in
(p = 0.11). This change in treatment effect was driven by a greater negative correction in pretreatment LCI than post-treatment LCI by both Lundin (pre-treatment correction -0.

Discussion
In agreement with previous studies, these data suggest that excretion of N 2 from body tissues affects MBW N2 outcomes. The effects of tissue N 2 are greater in patients with longer washouts. This contribution of tissue N 2 to FRC N2 and LCI N2 is less pronounced at earlier cut-offs of the washout. Application of correction equations for tissue N 2 significantly reduced, but did not completely eliminate, the effect of tissue N 2 on MBW N2 outcomes. Importantly, application of these tissue N 2 correction equations did not significantly alter treatment effects previously observed in interventional trials. Thus, while the excretion of tissue N 2 has a measurable effect on MBW N2 outcomes, correction for tissue N 2 using currently available approaches cannot be recommended at the present time.
FRC is an integral component of the calculation of LCI by MBW and therefore a reliable FRC is required to derive a reliable LCI. While there is no gold standard for the determination of FRC, body plethysmography and inert gas washout are the most commonly used techniques [16,17]. In the current study, FRC N2 was compared to FRC pleth and FRC SF6 to estimate the contribution of tissue N 2 excretion. With FRC pleth, the volume of all compressible intrathoracic gas is measured whereas only the volume of communicating lung units is measured with FRC N2 . Therefore, FRC measured by gas-dilution technique (such as MBW N2 ) should be equal to or less than that measured by plethysmography in the absence of endogenous production of the tracer gas [17]. FRC SF6 is also calculated using a gas-dilution technique, and because it is an exogenous, biologically inert gas that does not dissolve significantly in blood or other tissues, it was used as comparator to assess for the contribution of tissue N 2 excretion to FRC N2 .
We found that FRC N2 was systematically overestimated compared to FRC SF6 and, to a more variable extent, FRC pleth (Figs 2 and 5). This suggests that there is a systematic difference between these tests and that the observed differences were not entirely due to intrinsic differences between the MBW and plethysmographic techniques. While our analyses focused on the potential effect of tissue nitrogen excretion on this overestimation, there are other explanations for this disparity that could contribute to the observed differences that were not assessed in the current study, such as testing order, technical inconsistencies in the MBW equipment, and physical differences between SF 6 and N 2 tracer gases. Relationship between the contribution of tissue N 2 to LCI N2 and length of washout calculated at a) the traditional 2.5% washout cut-off and b) the 5% washout cut-off. The contribution of tissue N 2 to LCI N2 calculated at the 2.5% cut-off (LCI uncorr −LCI Cournand ) increased as washout time increased. However, this relationship was no longer observed at the earlier 5% cut-off.
https://doi.org/10.1371/journal.pone.0185553.g006 Table 5. Effect of applying Lundin and Cournand correction equations to previously published observational MBW data. Data are shown as pretreatment and post-treatment LCI with paired treatment effect. Values are presented as mean (SD) unless otherwise indicated.

Pre-treatment LCI
Post-treatment LCI Treatment effect mean difference (95%CI)

Ivacaftor [14]
Uncorrected 13.7 (3.7) 11.6 (4.1) -2. Correcting for tissue nitrogen in MBW measurements The order of tests could have inadvertently biased the results through effects of tissue hysteresis or other unknown mechanisms. In the original study, all plethysmographic testing was performed after the MBW testing and the order of MBW SF6 and MBW N2 was randomized [2]. All MBW-based outcomes can be affected by errors in gas concentration measurement, flowgas signal alignment, dead-space correction and other device-specific settings [19][20][21]. In this study, we used working-group recommended equipment and software settings on both the Exhalyzer D and AMIS 2000 devices and applied standardized quality control criteria to each MBW trial. We also confirmed the accuracy of the N 2 concentration calculation (as FN 2 is derived from measured O2 and CO 2 concentrations using the Exhalyzer device) across a range of gas standards. Despite our attempts to minimize technical software or device-specific inconsistencies, these cannot be completely ruled out as sources of systematic error that could contribute to the discrepancies observed.
The intrinsic properties of MBW SF6 and MBW N2 tests could also have contributed to these differences. The molecular properties of SF 6 and N 2 likely result in differences in their diffusion-convection fronts, which could potentially impact MBW outcomes [22]. MBW SF6 requires a wash-in equilibration phase as SF 6 is an exogenous tracer gas, and while standardized quality control techniques were implemented to attempt to ensure complete SF 6 washing, it is possible that incomplete wash-in of the SF 6 could result in altered excretion kinetics. Finally, the 100% oxygen washout phase in MBW N2 could also theoretically have pro-atelectatic effects, thus altering pulmonary gas flow dynamics. While simultaneous direct measurements of N 2 and SF 6 on the same device would permit an ideal comparison of these two MBW systems, unfortunately, high O 2 concentration impairs the ability of the AMIS 2000 respiratory mass spectrometer to measure N 2 concentrations and can therefore not be used to measure the two gases in the context of a 100% oxygen washout. Overall, our results need to be interpreted in the context of these potential limitations; nevertheless, the consistent overestimation of FRC N2 when compared to FRC pleth and FRC SF6 suggests that tissue N 2 likely contributes to this phenomenon.
Both FRC N2 and LCI N2 decreased significantly upon application of the tissue N 2 excretion equations in both healthy subjects and subjects with CF, with greater differences observed in CF. The estimates of the contribution of tissue N 2 to FRC N2 and LCI N2 are similar to those previously predicted by a two-compartment lung model including variable ventilation heterogeneity and dead space effects [3]. However, the difference between FRC SF6 and FRC N2 was significantly greater than the degree of correction applied by either Lundin or Cournand equations (Fig 3). Also, application of the correction equations only decreased the time-dependentrise in FRC N2 by~30% (Fig 4; Table 4). These findings suggest either that the equations used in this study underestimate the amount of tissue N 2 excretion, or that there are other factors in addition to tissue N 2 secretion that are driving this difference.
The Lundin tissue N 2 excretion equation is based on the average of measurements derived from healthy adults, therefore its application to MBW N2 data derived from subjects of varying size is limited. Compared to the Cournand equation, which was derived from subjects ranging from 9 to 44 years old and adjusts for a subject's body size, the Lundin equation may overestimate the effect of tissue N 2 excretion in smaller pediatric subjects. Although the Cournand equation may introduce less error overall in MBW N2 measurements from subjects with a range of body size, it assumes a constant rate of N 2 excretion from the body tissue which is unlikely to be the case in subjects of varying body composition and between health and disease. In a recently published study [23], the rate of tissue N2 excretion was simultaneously performed on MBWN2 and MBWSF6 washouts and confirmed the time-dependent nature of tissue N2 excretion and demonstrated higher rates of tissue N2 excretion during moderate exercise." Ideally, direct measurement of pulmonary N 2 excretion of tissue N 2 with modern equipment across a range of ages, body compositions and disease states would allow us to generate an optimal correction equation. However, due to the long duration of the testing and uncomfortable testing setup, replications of these early studies would be extremely challenging to conduct today, especially in children [6,9]. Furthermore, no mathematical correction for tissue N 2 excretions will be ideal for several reasons. First, even with modern technology, it is impossible to precisely isolate all of the N 2 in the lungs that was excreted from the body tissue, especially during the beginning of the washout when the relative proportion is very small; the derived equations are reflections of this imprecision. Second, the contribution of N 2 from the body tissue is likely dependent not only on time and body size, but also on factors such as cardiac output, tissue perfusion, body fat content, ventilation homogeneity, and dead space [3,6,8,[23][24][25][26]. Any number of these physiological factors could be altered in a disease like CF and could confound the estimation of tissue N2 excretion.
The extent to which the MBW N2 outcomes diverged from both MBW SF6 and MBW pleth was related to the length of the washout. This correlation makes intuitive sense, since individuals with longer washouts (greater ventilation inhomogeneity) spend a longer time at lower endtidal N 2 concentrations, thereby accentuating the relative contribution of excreted tissue nitrogen. Given this finding, we showed that the contribution of tissue N 2 can be minimized by calculating MBW N2 outcomes at earlier cut-offs of the washout, such as at the 5% normalized end tidal concentration of N 2 . Using an earlier cut-off of the washout has the additional benefit of shortening the total time it takes to perform an MBW test; however, there is some evidence that there may be a trade-off with decreased sensitivity to treatment efficacy [18]. Nevertheless, the use earlier cut-off for MBW N2 did not affect the significance of treatment effects observed in a study of Ivacaftor treatment [14], suggesting that the sensitivity of an MBW cut-off may depend upon the effect size of the intervention. The optimal MBW N2 cutoff for interventional studies may depend on study design and treatment.
Finally, to address the practical question of whether or not the correction for tissue N 2 excretion could affect the results of previously reported interventional studies, we applied tissue N 2 correction equations to raw MBW N2 data from a study that assessed the effect of ivacaftor on LCI [14]. Overall, this study had a large treatment effect (-2.2 LCI units) and we found that applying tissue N 2 correction equations attenuated the treatment response, but did not change the significance or direction of the treatment effect. This attenuation of the treatment response occurred primarily by reducing the post-treatment LCI by a greater amount than the pre-treatment LCI and is likely a reflection of the observation that tissue N 2 has a greater contribution in longer washouts. Taken together, these results suggest that non-correction for tissue N 2 release may result in marginally overestimated treatment effects. While this does not significantly affect the results of the studied trial, it is conceivable that smaller treatment effects could be amplified by non-correction for tissue N 2 .
In conclusion, MBW N2 outcomes are systematically different from MBW SF6 and plethysmography. We show that correction for tissue N 2 excretion using previously derived equations can reduce, but not eliminate, these differences. This suggests that either there are other physiologic/experimental factors contributing to this difference, or that the correction equations that were used underestimate the quantity of tissue N 2 excretion. Given our data, we suggest that there is currently inadequate knowledge of the true rate of pulmonary tissue nitrogen excretion to suggest a standard correction equation for this phenomenon in the calculation of MBW outcomes. Further study (ideally simultaneous MBW N2 and MBW SF6 measurements using an appropriately tuned mass spectrometer) could elucidate the contribution of tissue N 2 to MBW N2 outcome measures. Until this is clarified, it should be recognized that the magnitude of treatment responses measured with MBW N2 may be over-estimated by tissue N 2 excretion, however, application of correction equations in this study did not change the direction or significance of the treatment effects of a previously studied intervention.