To the Editors:
Considerable confusion exists about the clinical use of exhaled nitric oxide measurement in general, and its bronchial and alveolar contributions in particular, for instance in response to treatment. An additional effect needs to be factored in when considering the degree of alveolar nitric oxide abnormality and its response to therapeutic interventions that may or may not be targeted to the lung periphery. Indeed, the alveolar nitric oxide value computed from exhaled nitric oxide measurement at multiple flows with the so-called slope-intercept method [1, 2] can overestimate the true nitric oxide produced by inflammation in the alveolar air spaces. Such overestimation arises when the bronchial nitric oxide back-diffuses into the alveolar air space and thus contaminates the alveolar nitric oxide measurement with nitric oxide that originates from the more proximal airways. Two correction formulas have been published independently [3, 4] proposing to estimate true alveolar nitric oxide by subtracting from the measured alveolar nitric oxide a bronchial nitric oxide-dependent portion corresponding to back-diffusion. However, it has also been shown that airway constriction of peripheral conductive airways may at least partly impair back-diffusion [5]. Thus, in the case of peripheral lung disease, the application of correction formulas that assume unimpaired back-diffusion can erroneously lead to overcorrection and, ultimately, to negative alveolar nitric oxide values. The problem with the real lung is that it is difficult to judge whether and to what extent back-diffusion is impaired, although independent measures of small airway constriction could be envisioned in an attempt to determine this. In the meantime, we advocate here a more pragmatic approach.
One way to inspect uncorrected alveolar nitric oxide concentration (CA,NO) for true abnormality is by first plotting it against maximal bronchial nitric oxide production (J′aw,NO) as in figure 1a for data retrieved from 30 publications reporting both alveolar nitric oxide and bronchial nitric oxide production in asthma patients; if available from these asthma studies, data on normal control subjects were also retrieved (table 1). Each data point in figure 1a represents the uncorrected values of J′aw,NO and CA,NO corresponding to any given group of asthma patients or normal subjects retrieved from each study. In those papers where exhaled nitric oxide fraction at 50 mL·s−1 (FeNO,0.05) was reported instead of J′aw,NO, the latter was computed using the average multiplicative factor between J′aw,NO and FeNO,0.05 obtained from 15 out of the 30 papers where both were reported (mean±sd factor 45±4). From figure 1a, we can now assess each (J′aw,NO, CA,NO) data point with respect to a previously established “zone of normality” (dashed lines), which delimits combinations of CA,NO and J′aw,NO for which true alveolar nitric oxide is in fact normal, and any elevated CA,NO value can be attributed entirely to the increased J′aw,NO when full back-diffusion applies [4]. The 95% confidence interval around the regression line was previously obtained from experimental data on normal subjects and stable asthma patients [4]. Importantly, the experimental regression line itself (not represented here for clarity) corresponding to an average 1.7-ppb increase in CA,NO for every 1,000 pL·s−1 increase in J′aw,NO, was almost indistinguishable from that predicted by simulations of convective and diffusive gas transport in a lung model with normal peripheral airways [4].
It can be seen from figure 1a that, despite methodological differences and anthropometric variability in geographical regions from which the different research papers originate, almost all normal data fell within the zone of normality, and more specifically in the lower range of both J′aw,NO (<1,200 pL·s−1) and CA,NO (<5 ppb). At the other end of the spectrum (J′aw,NO >2,000 pL·s−1 and outside the zone of normality), we mostly observed data points for groups of asthma patients who were either steroid-naïve (open triangles) or in exacerbation (closed triangles). Two subsets of asthma patients warrant particular attention. On one hand, the patient groups with combinations of J′aw,NO and CA,NO located above the zone of normality are patients with a true increase of nitric oxide originating in the alveolar spaces, i.e. an alveolar nitric oxide in excess of what could be expected on basis of their corresponding bronchial nitric oxide production, even in case of full back-diffusion. In these patients, CA,NO values following a full back-diffusion correction would still be abnormal. On the other hand, patient groups with combinations of J′aw,NO and CA,NO below the zone of normality corresponded to patients for whom back-diffusion is hampered by considerable airway constriction, in which case back-diffusion correction would lead to negative alveolar nitric oxide values [19].
Once abnormality of the uncorrected CA,NO is established based on a J′aw,NO/CA,NO plot, the same representation can also help interpret response to treatment. Figure 1b shows data from 15 interventional studies in asthma patients [6–21] including five studies (solid symbols and lines) where two treatment arms were compared, either within the same group of asthma patients or between two comparable patient groups [6–11]. It can be seen that the majority of studies follows the “normal” CA,NO decrease with respect to J′aw,NO decrease, in which case the treatment effectively lowers bronchial nitric oxide production but does not really affect alveolar nitric oxide when taking into account the back-diffusion effect. Some studies follow a steeper than normal CA,NO decrease with J′aw,NO decrease, indicating a true alveolar effect, while others show a marked J′aw,NO decrease with no concomitant CA,NO decrease, usually in patients with pre-treatment data points located below the zone of normality. In fact, it has been shown recently that after steroid treatment, such patients may even paradoxically increase their uncorrected CA,NO value, probably because impairment to back-diffusion is lifted [19]. Surely, some interlaboratory methodological issues could influence the absolute value of what constitutes a “normal” slope in the relationship between CA,NO and J′aw,NO. However, when comparing slopes of CA,NO versus J′aw,NO between different treatment arms studied in the same laboratory, a relatively steeper slope should signal a more peripheral effect. In fact, for the five comparative studies (solid lines in fig. 1b), the relatively steeper slope did correspond to the treatment arm with a intended more peripheral therapeutic effect.
In summary, we have taken the opportunity to consider some of the issues with exhaled nitric oxide measurement that have frequently frustrated researchers willing to incorporate this biomarker of inflammation in their study protocol. It would not be the first simple, noninvasive test that has been characterised by an initial outburst of enthusiasm, followed by sound scepticism or discouragement because its interpretation proves to be more complicated than the test itself, at which point it becomes at risk of being all but abandoned. With the comprehensive compilation and interpretation of published alveolar nitric oxide data in asthma to date, we have attempted to reinforce the interest in the exhaled nitric oxide test and, in particular, the components representing alveolar and bronchial nitric oxide. We propose that before readily applying a full back-diffusion correction, individual values of CA,NO would be plotted versus their corresponding value of J′aw,NO, and assessed with respect to what has been previously obtained (fig. 1). Besides offering the possibility to diagnose possible equipment-related biases, such an approach could identify patients below the zone of normality for whom the full back-diffusion correction should not be applied. Finally, the proposed J′aw,NO/CA,NO data plots enable a direct comparison of different treatment interventions and identification of a more peripheral effect.
Footnotes
Statement of Interest
None declared.
- ©ERS 2012