© 2007 American Thoracic Society doi: 10.1164/rccm.200702-246ED
Exhaled NO: Forward, Backward, or Sideways?Dunedin School of Medicine, University of Otago, Dunedin, New Zealand Asthma is increasingly recognized as a syndrome comprising a number of inflammatory phenotypes. As such, it is not surprising that significant heterogeneity is also found in relation to treatment response. To date, clinicians have relied on assessing symptoms and/or changes in airway reactivity to evaluate the need for, or the response to, antiinflammatory treatment. But this is problematic: correlations between clinical parameters and airway inflammation are notoriously weak (1). To be rational, decisions regarding antiinflammatory therapy ought to be based on measuring airway inflammation. The advent of so-called "inflammometry" using induced sputum analysis or exhaled nitric oxide (FENO) represents an important advance in this regard. Using these methods particularly to distinguish eosinophilic from noneosinophilic inflammation is relevant because the former is associated with a positive response to corticosteroid therapy (2). FENO measurements, a surrogate marker for eosinophilic airway inflammation, are highly predictive of steroid-responsiveness in patients with airways-related symptoms (3), and levels change in relation to an increase or decrease of inhaled corticosteroid (ICS) dose (4). Against this background, two studies confirmed that a treatment algorithm based on induced sputum eosinophil counts results in significant improvements in asthma control compared with a strategy guided by symptoms and lung function (5, 6). Given that FENO measurements are much easier to obtain, it was obvious that the same question should also be addressed using FENO-based algorithms. In the latest study, reported in this issue of the Journal (pp. 231–237), Shaw and coworkers inform that, unlike the use of sputum cell counts, they were unable to identify significant benefits using FENO measurements as a guide ICS treatment, at least in terms of a reduction in exacerbations (7). The authors concede that their study was underpowered. But their results are supported by the absence of any significant trend toward improved asthma control, as measured by lower asthma control scores in the FENO group. This outcome, together with those reported in the two other studies to date (8, 9), provides us with a mixed salad. Although the study by Smith and colleagues was appropriately powered, the exacerbation frequency was lower, but not significantly, in the FENO strategy group (8). Interestingly in that study, the maintenance ICS dose was significantly reduced in the FENO group. The study by Pijnenburg and coworkers was not powered to demonstrate differences in exacerbation rates, but showed that airway hyperresponsiveness (AHR) was reduced by 2.5 doubling doses in the FENO strategy group compared with 1.1 doubling doses in the "symptoms" group (9). One might speculate that, if sustained, an improvement in AHR of this magnitude ought to be reflected in a comparably greater reduction in the exacerbation frequency over time. However, in the study by Shaw and colleagues, the improvement in AHR in the FENO group was only transient. Intuitively, one would have expected that, notwithstanding any design differences or weaknesses in the three studies, outcomes similar to those obtained using sputum cell counts to guide treatment would have emerged more consistently (if not statistically significantly) if benefits were going to be achieved using a FENO-based algorithm. Overall, a conservative interpretation of the three studies would suggest that the case for using FENO in this particular setting remains at best "not proven."
The reasons for the apparent discrepancies between the studies are worth closer examination. First, there is the issue of study design. In the studies by Smith and coworkers (8) and by Shaw and colleagues (7), a two-tier algorithm with a single FENO cut-point (35 ppb and 26 ppb, respectively) was used to determine whether or not the ICS dose ought to be changed. In the study by Pijnenburg and coworkers (9), a three-tier algorithm including a middle tier, defined by a low FENO (< 30 ppb) and the presence of ongoing symptoms, was used to determine "no treatment change." In devising such algorithms, the choice of cut points is critical. The lower the FENO cut-point, the higher will be the mean ICS dose requirement and vice versa, with consequent differences in the resulting exacerbation rate. But the importance of cut-points applies equally to the control groups, a point which has not been frequently considered. The thresholds at which symptoms and peak flows are set to prompt treatment change in the control group will also influence study outcomes. Shaw and colleagues sought to address this issue by choosing a previously validated ACQ score of 1.57 as the chosen cut-point, but the net potential for study bias is difficult to assess. Second, there is the effect of concurrent long-acting There is also the important issue of discordance between FENO levels and simultaneous sputum eosinophil counts. The proportion of false-positive FENOs recorded by Shaw and coworkers (i.e., where FENO levels did not reflect the true presence of eosinophilic airway inflammation) was high. This was noted to occur in over 50% of those cases in which, despite taking the maximum permitted ICS dose, FENO levels remained > 26 ppb and sputum was simultaneously obtained. This issue has previously been highlighted by Pijnenburg and colleagues (10). Unlike sputum eosinophils, a zero value for FENO cannot be achieved even with optimum anti-inflammatory treatment. In some patients, levels remain resolutely high. The reasons for this are complex, but include the fact that the "normal range" for FENO is influenced by constitutional as much as by environmental and pathophysiologic factors, and is skewed to the right. This perhaps explains why FENO measurements appear to be less successful than induced sputum cell counts to guide ICS dosing, particularly when they fall in the range of 25 to 50 ppb. These issues are indirectly highlighted in a second article in this issue (11). Travers and coworkers (pp. 238–242) provide reference values in relation to a range of common factors known to affect FENO, including atopy, current cigarette smoking, and sex. In a recent larger population study, Olin and colleagues (12) reported that, after adjusting for height (an assumed surrogate for airway epithelial surface area from which exhaled NO is derived), the effect of sex on FENO is not significant. In fact, the weight of evidence suggests that gender is a significant issue (13–15). Given the magnitude of the differences between males and females reported by Travers and coworkers (averaging 25%), this effect is clearly something that now needs to be carefully considered. In practice, reference values obtained in healthy subjects have a very limited role when interpreting sequential FENO results in patients with diagnosed asthma, such as were studied by Shaw and colleagues. Nevertheless, if differences in FENO related to factors such as sex are consistent, then perhaps the analysis and interpretation of algorithm-based studies may have to take them into account. Cut-points for males may not be the same as cut-points for females. Inevitably, there are yearned-for simplicities and high expectations for a new test such as FENO which are unfulfilled. Clearly, FENO is not to the management of asthma what the sphygmomanometer is to hypertension, nor are "normal ranges" strictly comparable to those which apply in spirometry. But clarifying the scope and interpretation of FENO measurements remains a highly relevant research endeavor. Occasionally, the way forward may seem like a step backward...or perhaps a step sideways? Whatever our perception, it is surely a step in the right direction. FOOTNOTES Conflict of Interest Statement: D.R.T. has received lecture fees and financial support for a research project ($20,000) from Aerocrine, Sweden, manufacturer of nitric oxide analysers. REFERENCES
Related articles in AJRCCM:
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
-agonist use (LABAs). LABAs reduce the frequency of exacerbations, but where this is the primary end-point, permitting their use means that a much larger study size is required. The use of LABAs may be "real-life," but if the objective is to determine the impact of an intervention on exacerbation rates, then the confounding effect of LABA use should be eliminated. 
