Corticosteroid Therapy in Acute Respiratory Distress Syndrome: Title and subTitle BreakBetter Late Than Never?

The annual incidence of acute respiratory distress syndrome (ARDS) has been estimated at 150000 cases in the United States.¹ Although its outcome may have improved in the past 2 decades,^{2 - 3} ARDS remains associated with a very high mortality, ranging from 35% to 65% of affected patients.^{2 ,4} Severe hypoxemia (ie, an oxygenation ratio of PaO₂ to fraction of inspired oxygen [FIO₂] of <200) is the hallmark of the syndrome,¹ but most fatalities result from associated conditions, especially multiple organ failure and sepsis, and less commonly from intractable respiratory failure.²

Acute lung injury and ARDS are characterized by increased permeability of the alveolocapillary membrane, whether caused by direct lung injury, such as pulmonary infection, or indirect lung injury, such as observed in severe sepsis or a host of other nonpulmonary conditions. Extensive research efforts have explored the pathophysiology of cellular pathways and biochemical derangements occurring in ARDS. Early ARDS is associated with a local and systemic inflammatory response, with high levels of inflammatory mediators in bronchoalveolar lavage fluid.⁵ Soon after acute lung injury, lung repair and remodeling occurs, associated with high levels of growth factors and markers of collagen synthesis in bronchoalveolar lavage fluid, and possible evolution toward fibrosing alveolitis.⁵

Experimental studies suggest that lung injury may be aggravated by high alveolar pressure used during mechanical ventilation to maintain alveolar ventilation and oxygenation,⁶ applied to a progressively smaller lung area available for gas exchange.⁷ Based on these observations and recent clinical reports,⁸ a consensus is emerging on the optimal strategy for delivering mechanical ventilatory support in the early phase of the syndrome.⁹ Attempts to pharmacologically blunt the early inflammatory response thought to contribute to injury of the alveolocapillary membrane have targeted various steps of the inflammatory network and have used corticosteroids, anticytokine therapy, antioxidants, cyclooxygenase inhibitors, or prostaglandins. Unfortunately, none of these pharmacological approaches has been established as both safe and effective⁴ ; as for severe sepsis and septic shock,¹⁰ results of randomized trials of early high-dose corticosteroids in ARDS also have been disappointing.^{11 - 12}

In the randomized trial reported in this issue of THE JOURNAL, Meduri et al¹³ have targeted the secondary active fibrosing process associated with postaggressive alveolar remodeling and tested the efficacy of high-dose and prolonged methylprednisolone therapy in unresolving ARDS. This placebo-controlled trial, conducted in 4 centers, enrolled 24 high-risk patients in whom the lung injury score (a composite score including oxygenation, extent of radiographic involvement, positive end expiratory pressure level, and thoracic compliance)¹⁴ had not improved after 1 week of mechanical ventilation. The results appear promising, with a marked reduction (12% vs 62%, P=.04 using the Fisher exact test) in hospital mortality for methylprednisolone recipients.

There are several challenges associated with the interpretation of this trial, related to its small sample size and to its design, including a planned crossover of patients who did not respond by day 10 to the initially allocated treatment. Although this approach reflects a belief that every patient should be offered methylprednisolone, it complicates interpretation of the results. Randomization of treated and control subjects was in a 2:1 ratio, so that 16 patients were allocated to methylprednisolone and only 8 to placebo (4 of whom subsequently crossed over to methylprednisolone with only 1 survivor). A concern with half the placebo recipients crossing over therapy is the potential risk of intensive care unit–acquired infection. Indeed, 75% of all trial patients acquired infection, and the incidence rate of infection in methylprednisolone recipients was almost twice as high (risk ratio, 1.8; 95% confidence interval, 0.86-3.75) as in placebo recipients.

Enrollment of only 24 patients in a clinical trial widens the confidence interval around the effect estimate and makes imbalance of prognostic factors between groups a major threat, even if the trial is randomized. In this instance, the absolute difference in hospital mortality between treated and control subjects may range anywhere between 11.5% and 88.5%. Although treated and control subjects had similar characteristics at onset of ARDS, the former group had lower sepsis, lung injury, and organ dysfunction scores at study entry 10 days later, all pointing to a lesser severity at enrollment and to a larger improvement in the prerandomization period. Whereas none of these differences taken individually was statistically significant, adjustment using a global severity score at study entry to handle these subtle imbalances in prognostic factors would have been useful. Given these limitations, the beneficial effect of steroids reported by Meduri et al¹³ must be interpreted with caution, and these results should be confirmed in a larger trial before late administration of steroid therapy gains widespread acceptance in unresolving ARDS.

Nevertheless, this trial provides a strong impetus for further testing of a new approach in an area that needs innovative therapeutic strategies. However, only a small fraction of patients with ARDS (ie, those patients surviving the first week of the syndrome but who failed to improve substantially, who appear to have persistent lung inflammation, and who do not have uncontrolled infection) would be candidates for late steroid therapy. Central to such patient selection is better characterization of lack of clinical improvement rather than use of only the lung injury score. Aggressive search for and treatment of infectious complications is also necessary.¹⁵ Comprehensive and repeated surveillance for infection during therapy has been emphasized elsewhere by Meduri et al.¹⁵

Several questions remain, pertaining to the timing, dosage, and duration of late steroid therapy in ARDS. Based on their prior experience, Meduri et al¹⁵ used a methylprednisolone dosage of 2 mg/kg for the first 2 weeks, tapered to 1 and 0.5 mg/kg per day during the following 2 weeks.¹³ Although the authors make a convincing case for maintaining methylprednisolone therapy for more than 10 days, it is unknown whether this regimen is optimal. Similarly, the appropriate time window for corticosteroid administration, between early acute injury and established postaggressive fibrosis, cannot be derived from this study. Based on the poor outcome in patients secondarily switched to methylprednisolone, the authors suggest that corticosteroids given at an intermediate stage of evolution of ARDS have a fairly narrow therapeutic window (ie, between 7 and 15 days after onset of ARDS). Prior uncontrolled experience by the same group¹⁵ suggests that patients may improve even when corticosteroids are administered more than 15 days after onset of ARDS.

These questions should be kept in mind when designing future trials evaluating treatment of ARDS, which are needed to confirm or refute the findings of Meduri et al.¹³ Meanwhile, these investigators should be commended for having performed an innovative and thought-provoking trial and for offering some prospects for further improved outcome for this complex and deadly syndrome.