Mitigating increased variability of multiple breath washout indices due to tidal breathing

To the Editor:

Multiple breath washout (MBW) tests have gained interest for clinical use, with normal values being published for the lung clearance index (LCI) in the paediatric [1, 2] and adult range [3, 4]. In adults, the greater interest in mechanisms of acinar and conductive ventilation heterogeneity (reflected in S_acin and S_cond, respectively) drove the adoption of a 1 L breathing protocol to ensure a clear estimate of phase III slope. Alternatively, natural breathing protocols aimed to enhance feasibility in paediatric settings [5] have also been used in some adult studies [6]. Since S_acin or S_cond are indirect measures of lung structure and its potential abnormality in disease [7, 8], the manoeuvre (natural breathing or not) by which these indices are obtained does not directly alter the underlying process, but merely affects the magnitude of its estimates. In specific age groups, or in patient groups of any age with severe lung disease, a 1 L tidal volume (V_T) may not be achievable or may result in an altered functional residual capacity (FRC). V_T increases from 0.5 to 1.3 L have led to FRC decrease by 17% in children [9] and by 7% in adults [10]; corresponding LCI increases were observed in children, but not in adults. In natural breathing studies, S_acin and S_cond are usually compensated by V_T multiplication to account for differences in lung size (multiplication by FRC) and in breathing pattern (multiplication by V_T/FRC) [5]. In the present work we systematically studied the effect of natural tidal breathing (MBW_nat), 1 L breathing (MBW_1L) and 1.5 L breathing (MBW_1.5L) on MBW indices LCI, S_acin, S_cond and the V_T compensation of the latter two, with an aim to distil a breathing modality that may enhance clinical utility of MBW in the future.

40 healthy young adults (20 males, 20 females) were recruited (local ethics committee BUN143201836071) and all nitrogen MBW testing was performed as previously described [11]. Subjects performed nine MBW tests (three sets of three trials, always performing the first set during natural tidal breathing, to avoid a potential impact on this from any larger V_T breathing before it). Friedman test and multiple regression were performed using MedCalc (version16.4.3, Mariakerke, Belgium).

In the study cohort (mean±sd age 24.4±3.4 years), mean±sd z-scores for FRC, LCI, S_acin and S_cond obtained with the MBW_1L test were 0.1±0.8, −0.1±1.1, −0.1±0.8 and −0.1±1.7, respectively, using equipment-specific reference equations [11]; Global Lung Function Initiative-based z-score for forced expiratory volume in 1 s was 0.3±0.09 [12]. Mean±sd trial durations were 191±41 s for MBW_nat, 124±26 s for MBW_1L and 98±21 s for MBW_1.5L. Figure 1 illustrates the effects of V_T on raw value FRC, LCI, S_acin and S_cond. Compared to MBW_1L, natural tidal breathing induced statistically significant but small LCI increases, and a considerable S_acin increase, on average from 0.070 L⁻¹ (MBW_1L) to 0.255 L⁻¹ (MBW_nat). When V_T-compensated as per guidelines [5], a significant difference persisted but mean S_acin·V_T increased relatively less, from 0.077 (MBW_1L) to 0.150 (MBW_nat). While S_cond did not show a significant difference (MBW_1L 0.030 L⁻¹ versus MBW_nat 0.029 L⁻¹), it reached significance for S_cond·V_T (MBW_1L 0.033 versus MBW_nat 0.017), suggesting overcompensation by V_T. Interestingly, V_T compensation works better between MBW_1L and MBW_1.5L, where V_T multiplication better neutralises the consistent S_acin and S_cond decreases observed with increased V_T (mean S_acin·V_T 0.069 for MBW_1.5L versus 0.077 for MBW_1L, and S_cond·V_T 0.036 for MBW_1.5L versus 0.033 for MBW_1L).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1

Multiple breath washout (MBW)-derived indices as a function of tidal volume (V_T) for male (circles) and female (triangles) normal subjects. Closed symbols are mean±sd and positioned at mean V_T values corresponding to target volumes of respectively MBW_nat, MBW_1L and MBW_1.5L. Open symbols are individual data points versus individual V_T values, connecting MBW_nat, MBW_1L and MBW_1.5L data sets for any given subject (dotted lines). a) Functional residual capacity (FRC); b) lung clearance index (LCI); c) acinar ventilation heterogeneity (S_acin); d) conductive ventilation heterogeneity (S_cond).

To investigate whether the experimentally observed degree of V_T dependence of S_acin could be replicated based on the underlying model of diffusion–convection interaction, simulations were performed with an adult lung model recently used for simulation of S_acin increases typically seen in COPD [7]. With the model in its normal baseline state (and simulated FRC of 3000 mL), this obtained simulated S_acin values of 0.134 L⁻¹ (750 mL), 0.085 L⁻¹ (1000 mL), 0.060 L⁻¹ (1500 mL); with V_T compensation, corresponding simulated S_acin·V_T values were 0.101, 0.085 and 0.090.

A striking V_T effect in figure 1 is its impact on inter-subject variability. While inter-subject coefficient of variation for this young adult cohort was similar across the studied V_T range for FRC (MBW_nat 23%; MBW_1L 25%; MBW_1.5L 25%) and LCI (MBW_nat 5.8%; MBW_1L 4.8%; MBW_1.5L 4.9%), this was not the case for S_cond (MBW_nat 147%; MBW_1L 37%; MBW_1.5L 32%) nor for S_acin (MBW_nat 59%; MBW_1L 30%; MBW_1.5L 36%). Also with V_T compensation, inter-subject S_acin·V_T and S_cond·V_T variability for MBW_nat were 58% and 128%, respectively, and only in the case of S_acin·V_T could this be partly accounted for by inter-subject variability in V_T and FRC (R²_adjusted=0.38; V_T: r_partial=−0.35, p=0.03; FRC: r_partial=0.64, p<0.001). What is also apparent from figure 1, is that for any individual with a given FRC, there is a steep dependence of S_acin on V_T near natural breathing, such that small variations in V_T can result in large S_acin variations. Hence, if subjects are allowed to freely use their natural breathing, instead of a weight- or height-based fixed tidal V_T, this will be detrimental to variability of study outcomes. Also, if a treatment were to increase natural V_T, a S_acin decrease could signal a treatment effect or a purely volumetric effect, or both.

We show here quantitatively, a V_T effect that has long been known to affect ventilation distribution, with increasing V_T generally decreasing phase III slopes [13, 14]. The V_T compensation is similar to what is done when comparing species, e.g. phase III slopes from humans (in L⁻¹) and rats (in mL⁻¹) [15], where both V_T and FRC are scaled by a factor 1000. The present experimental data show that the V_T compensation did attenuate dependency of both S_acin and S_cond on V_T in the 1.0−1.5 L V_T range, but that it could not fully compensate S_acin, and even overcompensated S_cond in the case of MBW_nat. In the case of S_acin, this was supported by model simulations where the purely volumetric effect on diffusion–convection interaction in the lung periphery could be assessed. By contrast, S_cond and the conductive ventilation heterogeneity portion of LCI, cannot be readily simulated unless complex patient-specific models are constructed [8].

Besides the increase of S_acin with natural tidal breathing, its variability also increases considerably. Whilst the limits of normal for healthy reference data will take care of this inherent variability, it is tempting to suggest that encouragement to achieve slightly deeper breaths than their natural breathing may benefit MBW measurement variability. S_acin values would indeed become smaller with greater V_T, but the V_T compensation would work better as one moves away from natural tidal V_T, as is seen here in the higher V_T range. The recent consensus statement suggested that when assessing S_acin and S_cond, “an initial V_T range of 10–15 mL·kg⁻¹ can be used but may need to be adjusted for the individual patient depending on the expirogram seen” (table 5 in [5]). Our study, where V_T was 9±2 mL·kg⁻¹ for MBW_nat and 16±3 mL·kg⁻¹ for MBW_1L, shows the benefit of the larger V_T in reducing S_acin and S_cond variability. Of note, the estimated population-based V_T ranges encountered in recent normative studies were ∼13 mL·kg⁻¹ in children [1] and ∼14 mL·kg⁻¹ in adults [4], suggesting that this phenomenon may be less problematic in that paediatric age range (6–18 years) and in existing adult data. As an additional benefit, our data show that the larger V_T reduces MBW test duration, i.e. time to achieve the end-of-test concentration threshold for LCI computation.

In conclusion, the quality control required for MBW indices is more complicated than for conventional lung function tests, partly because of V_T effects, which may be amplified during natural breathing. In interventional studies, alterations in breathing pattern should be scrutinised when interpreting reported changes in S_acin and S_cond, and identifying actual treatment effects. Adoption of procotols with a fixed V_T (or higher V_T range) rather than natural breathing in study populations where this is feasible, would be expected to decrease S_acin and S_cond variability, improve accuracy of any V_T compensation applied, and also shorten MBW test duration. LCI variability is less affected by V_T and is the more robust index, suggesting it may be more suitable across widely varying patient populations. Future research, within other age ranges and also in the setting of severe lung disease, are needed to examine the generalisability of these findings.

Shareable PDF