Long term oxygen therapy (LTOT) is one of the few treatments which has significant survival benefits in patients with severe hypoxaemia. It may modify disease progression, as indicated by a slower progression of hypoxia induced pulmonary hypertension,^1–4 and the acute reduction in pulmonary hypertension to oxygen administration has been suggested as predictive for the survival benefit in individual patients. Reduced pulmonary vascular resistance and hence the reduced load on the right heart is probably the most important working mechanism of LTOT. In less severe hypoxaemic patients the benefits of LTOT on survival are less clear.⁵ Other benefits of oxygen administration are generally accepted. Reduced ventilation, especially during exercise, helps to avoid dynamic hyperinflation and hence reduces symptoms and increases exercise tolerance in the majority of patients with chronic obstructive pulmonary disease (COPD), even in patients with mild hypoxaemia.⁶ There is also some evidence to support the suggestion that LTOT may improve cognitive function in hypoxaemic COPD patients⁷ and may improve health related quality of life.⁸

LTOT is therefore a recognised treatment in hypoxaemic patients⁹ and has been reimbursed in most healthcare systems. During exercise training oxygen supplements are administered to enhance training intensity¹⁰ or relieve symptoms.

Despite the proven benefits of oxygen therapy, researchers should remain critical towards interventions.¹¹ In this issue of Thorax Carpagnano et al¹² present interesting data that potentially invite us to refine our view on the benefits of oxygen therapy in COPD. The authors investigated the effects of acute administration of hyperoxia (Fio₂ 28%) on markers of oxidative stress and inflammation in exhaled breath condensate. They found that exposure to increased inspiratory oxygen fractions for 1 hour exacerbated 8-isoprostane and interleukin (IL)-6 concentrations (already raised breathing ambient air) compared with control subjects. Intriguingly, the effect of oxygen breathing was comparable between healthy subjects (IL-6 +68%, 8-isoprostane +79%) and patients with COPD (IL-6 +31%, 8-isoprostane +49%). In other words, the effects of oxygen breathing were not restricted to COPD. In addition, the increases in both markers were significantly interrelated.

Although the data by Carpagnano et al are tantalizing for researchers in this field, the clinical relevance of the findings is not yet clear. Firstly, the use of markers in exhaled air is not an easy technique and it is difficult to reproduce findings in other laboratories.^13,14 Secondly, the magnitude of the increase in IL-6 and 8-isoprostane is difficult to put into context. The same research group has already shown that, in smokers, IL-6 levels in exhaled air were more than double (+115%) those observed in non-smokers. In heavy smokers IL-6 was +184% above control levels. The changes observed with oxygen breathing (+68% in healthy subjects and +31% in COPD patients) are therefore relatively subtle—and the clinical relevance might also be.¹⁵ Thirdly, the authors studied only one time point (after 1 hour of oxygen breathing), which makes it difficult to extrapolate to LTOT. It would be useful to know whether the observed effects are transient or whether the increased oxidative stress and inflammatory markers remain. Lastly, oxygen administration may exert different effects in the lungs from the “periphery”. COPD is more and more recognised as a systemic disease¹⁶ or a disease with systemic consequences. Oxygen administration may protect against the systemic consequences of COPD. For instance, oxygen administration has been shown to protect against systemic oxidative stress during a bout of exercise¹⁷ and, interestingly, Carpagnano et al¹⁸ have confirmed elsewhere that temporary hypoxia induced, for instance, by sleep apnoea leads to an increase in the markers of oxidative stress. These are normalised when overall oxygenation is improved with continuous positive airway pressure (CPAP). In patients with severe gas exchange disturbances, hyperoxia in the alveolar spaces may be needed to guarantee relative normoxia in the periphery of these patients. It is generally recognised that tissue hypoxia contributes to weight loss through the activation of the NF-κB pathway, which activates an inflammatory cascade releasing IL-6 and tumour necrosis factor (TNF)-α¹⁹ leading to tissue wasting. Weight loss—especially the loss of lean tissue—is in itself is a negative prognostic factor and should be avoided in COPD.²⁰ Since the same NF-κB pathway has been suggested to play a role in hyperoxia induced oxidative stress, it is not clear at present whether increased Fio₂ is good or bad.

Hyperoxia up to concentrations of 80% Fio₂ did not seem to lead to weight loss in rats.²¹ In the MRC LTOT trial¹ patients surviving in the LTOT arm did not tend to lose weight, nor did their lung function deteriorate more rapidly than in the control arm. Hence, somewhat increased Fio₂ values used in LTOT or during exercise to improve tissue oxygenation are therefore probably clinically superior to normoxia in the lungs, leading to relative tissue hypoxia. Evidence of “harm” induced by relatively modest Fio₂ is absent. Oxygen therapy therefore remains recommended—if not necessary—in patients at risk of tissue hypoxia.

In summary, the study reported by Carpagnano et al may shed new light on the effect of clinical doses of pulsed oxygen therapy on patients with COPD, and could be interpreted as a potential sign to be cautious in using oxygen therapy in these patients as it may exacerbate rather than alleviate the bronchial inflammation by inducing hyperoxia induced oxidative stress.

This study invites further research rather than a change in clinical routine.