Steroids for ARDS James Beck Clinical Problem A 60 year old lady re-presented to ICU with respiratory failure. She had previously been admitted for fluid management and electrolyte correction having presented with severe dehydration and markedly abnormal potassium, sodium, phosphate and magnesium. She had a background of alcohol excess and chronic pancreatitis. She was a smoker of approximately 20 per day. She had a 2 month history of diarrhoea, which had been investigated and was felt to be secondary to pancreatitis. Her first ICU admission was brief, once euvolaemia and normalising electrolyte levels had been achieved she was discharged to the ward having never required vasopressor or ventilatory support. She was soon re-referred with type 1 respiratory failure. This was felt to be fluid overload but management for hospital acquired pneumonia had also been instigated. Management Initial management was with empirical antibiotics and spironolactone (in view of her ongoing hypokalaemia). She had Non Invasive Ventilation (NIV) for her respiratory failure. The course of antibiotics completed with no improvement. At one week into her second ICU admission she was still alternating between NIV and nasal hi-flow, with fractional inspired oxygen (FiO 2 ) concentrations of 0.5 to 0.75. Chest radiographs showed persistent bilateral airway shadowing consistent with pulmonary oedema. Diuretics were increased and a respiratory specialist opinion sought. The advice received was to further increase diuretics. Echocardiography was unremarkable. At approximately three weeks from admission, she had static oxygen and ventilatory support requirements, and was considered to be well diuresed. The clinical opinion was that this represented an established picture of Acute Respiratory Distress Syndrome (ARDS). A high resolution computed tomography (HRCT) scan was undertaken. This scan showed areas of bronchiectasis but was also consistent with severe ARDS. The degree of lung injury shown on scan, when considered alongside a clinical picture of a frail patient who had not improved despite prolonged ICU management, led to discussions with the patient about the appropriateness of escalation to intubation and ventilation if it was needed. She stated that this would not be something she wanted. It was decided, in conjunction with the respiratory
physicians, to use pulsed methylprednisolone as a final possible therapeutic intervention. She was given 500mg once daily for 3 days. Over the following week she made a gradual improvement and was discharged on 35% oxygen. She is still currently an inpatient, 14 days since ICU discharge, but is no longer requiring oxygen therapy. Discussion My understanding of the evidence for steroids in ARDS was limited, this led to the choice of subject for this expanded case summary. ARDS begins with a trigger pathology or insult. This can be direct, such as pneumonia or indirect such as pancreatitis. The pathophysiology is of a diffuse neutrophilic inflammatory lung disorder. This causes dysfunction of the alveolar-capillary membrane leading to pulmonary oedema. At the same time there is microvascular dysfunction causing ventilation-perfusion mismatching. The final process is a fibroproliferative one. The latest diagnostic criteria 1 for ARDS is the Berlin Definition: Acute, meaning onset over 1 week or less from a trigger Bilateral opacities consistent with pulmonary oedema must be present and may be detected on CT or chest radiograph PF ratio <300mmHg (40kPa) with a minimum of 5 cmh20 PEEP (or CPAP) Must not be fully explained by cardiac failure or fluid overload, in the physician s best estimation using available information an objective assessment (e.g. echocardiogram) should be performed in most cases if there is no clear cause such as trauma or sepsis. The patient in question clearly met these criteria. One of the challenges of analysing the trial data is that some older studies used differing diagnostic criteria. The preceding, and broadly similar, definition was the American-European consensus definition. One of the proposed therapies is high dose steroids. The intention is to dampen down the inflammatory response, there is a consequent possibility of immune suppression, secondary infection and neuromyopathy. There have been a number of trials with conflicting outcomes. A search of the literature produced five randomised controlled trials that studied therapeutic steroids for ARDS. Two of these trials were in persistent (>5 days) ARDS, these would correspond most closely with the described case. There have also been a number of cohort studies. Early ARDS
The most recent RCT in early ARDS was undertaken by Meduri and published in 2007 2. This trial used the American-European consensus definition of ARDS, was of medium size (91 patients) and was double blind and placebo controlled. Randomisation method was unclear. Patients were recruited and treated within 72 hours of diagnosis. The treatment group received 1mg/kg methylprednisolone loading dose followed by a decreasing dose infusion over 28 days. The treatment group showed improved respiratory function and oxygenation at day 7. The treatment group also showed an improved ICU mortality but not hospital mortality. There was not a significant increase in the infection rate. Overall a solidly conducted trial. Ventilation does not appear to have been conducted to the modern 6ml/kg tidal volume standard, the trial was well underway when the ARDS Network tidal volume trial 3 was published. The authors argued that the prolonged treatment duration was beneficial compared to previous studies. So in summary, improved oxygenation, respiratory function and ICU but not hospital mortality. The next study, by Annane et al 4, was in fact a post hoc analysis of a trial studying steroids in septic shock. This consequently bring into question the validity. The original trial was placebo controlled, randomised and double blind. Again randomisation method was unclear but it took place within 8 hours of the onset of shock. The authors selected 177 patients (from 300 enrolled in the parent study) with ARDS based on American-European consensus criteria. The treatment arm received hydrocortisone 50mg 6 hourly and fludrocortisone 50 microg once daily for 7 days. Short synacthen test (SST) non-responders with ARDS had a survival benefit in the treatment arm at 28 days and for ICU and hospital mortality. There was no difference between arms in patients who responded to the SST (small numbers of patients). The non-responder treatment group also showed improvements in oxygenation and ventilation. When responders and non responders are analysed together the statistical significance of the positive results disappears. Non-ARDS patients in the trial did not see any benefit in the treatment arm. Similar to Meduri, no increase in infectious complications was noted. This study, whilst not designed specifically to address this question, was robust in design. It should be noted again that ARDS patients were ventilated to an average of 9ml/kg tidal volume and therefore were not being managed in a modern fashion. Again we see improved oxygenation and respiratory function but also with improved ICU and hospital mortality in SST non-responders only. Despite the non-ards treatment group absence of benefit I think a question remains as to whether the benefit of steroids in this trial is in sepsis or in ARDS. A much older study by Bernard et al in 1987 5 enrolled 99 patients. Diagnostic criteria were not in line with modern definitions. The treatment arm received very large (30mg/kg) doses of methylprednisolone 6 hourly for 24 hours only. There was no difference in the two arms of the trial. Late ARDS
The ARDS network published their large study on steroids for ARDS in 2006 6. 180 patients with ARDS (American-European consensus) between 7 and 28 days from diagnosis were randomised via permuted blocks. The treatment arm received methylprednisolone 2mg/kg loading dose, followed by 0.5mg/kg six hourly for 14 days and then 0.5mg/kg 12 hourly for 7 days and then a tapering dose over either 2 or 4 days depending on patient condition. The trial was double blind and placebo controlled. Interestingly the ventilator strategy changed half way through recruitment due to the previously mentioned tidal volume study 3 by the same research group. There was no difference in 60 or 180 day mortality. The treatment group did have shorter periods of ventilation and improved oxygenation but similar ICU stay length. Severe neuromyopathy was significantly more likely in the treatment group but not infection. Perhaps associated with the neuromyopathy was an increased risk of return to ventilatory support in the treatment group. Sub-group analysis demonstrated that initiation of steroids two or more weeks from diagnosis was associated with increased mortality. This was a well designed study but the authors do acknowledge the significant changes in practice that occurred in intensive care medicine over the 7 years of recruitment, in theory those changes would be mirrored in both arms of the trial but it remains a potential source of error. In summary no significant benefit, some increased risk and potentially increased mortality if initiated late. This is the largest RCT in the area of study. Meduri s previous paper on this subject was published in 1998 7. This was a small (24 patient) study. Again randomised, double blind and placebo controlled. Patient selection was ARDS (American-European consensus) with no improvement at 7 days. Treatment was initally methylprednisolone 2mg/kg/d, then tapered dosing for 32 days. Four patients were blindly crossed from one arm to the other for failure to improve. Improvements in oxygenation and ventilation were seen in the treatment arm. Hospital and ICU mortality was also better in the treatment arm. Infectious complications were not increased. Late initiation of treatment (seen in the crossovers) was associated with a significantly increased failure rate. The authors argue that once progression to fibroproliferation occurs then benefit is substantially reduced. These are small numbers to make such assumptions on however. The authors also argue that prolonged course steroids are key to effect, this is based on theirs and others previous work, including animal models. This trial, whilst small, has been well executed with one proviso. They study authors stopped recruitment earlier than intended based upon sequential analysis of the results, this informing them that they had reached a statistically significant result for their primary endpoint. This potentially exaggerates the treatment effect. Again this study predated the 6ml/kg ventilatory strategy. In summary an improved mortality and oxygenation/ventilation. Again some concern about late initiation of steroids. But a small study with potentially exaggerated results. Cohort studies and Systematic Review
Formal systematic review and meta-analysis 8 reveals 10 cohort studies in addition to the RCT s described. All bar one showed no mortality benefit, the one trial showing a benefit was small. Meta-analysis showed no evidence of benefit or harm in terms of mortality and infection rates. Combining the cohort studies with the RCT s detailed above gives a total of 1474 patients enrolled. The authors of the systematic review highlight a concern that possible early gains in terms of ICU mortality are then reversed in trials with longer follow up, they postulate this is due to immune-suppression and neuromyopathy. This is certainly seen in the ARDS 2006 6 paper and the Meduri 2007 2 paper. Subgroup analysis of timing of steroids is interesting. Meta-analysis 8 of both RCT s and cohort studies shows a more convincing trend to mortality benefit in persistent ARDS (>5days RCT s - relative risk (RR) 0.52, 95% confidence intervals (CI) 0.11-2.52) than early (<3 days RCT s RR 0.86, CI 0.71-1.04). This must be viewed with suspicion however, with significant evidence of study heterogeneity and wide confidence intervals that include a relative risk of 1.0 for both groups. Sub group analysis by aetiology also brings into question varying response to steroids from differing aetiologies. Influenza induced ARDS appearing to have a statistically significant increased mortality with steroid management and post operative ARDS having a decreased mortality (both cohort studies only) 8. Lessons Learnt I see no convincing evidence to support the use of corticosteroids in ARDS currently. It does however merit further study. In some ways there is similarity to the argument in favour of proning. For years trials showed improved oxygenation but no mortality benefit but eventually a trial has managed to show reasonably convincing evidence of a mortality benefit 9. It may be that steroids work with the right patient population and the right timing. There may be a window somewhere between persistent ARDS and 14 days where treatment is beneficial, but the increased mortality beyond 14 days 6 is a concern. There may be some patient groups more likely to benefit or be harmed. Future trials should assess timing of steroids and follow up should be to at least 60 days. They should either focus on a particular patient group or be powered for adequate sub group analysis. I do not believe we have yet identified a most appropriate dosing regime or indeed formulation of steroid, this would also require further study. There are at least two trials currently recruiting to further study this area, one using dexamethasone (NCT01731795) and one using hydrocortisone (NCT01284452).
In the clinical vignette above the decision to try steroids was a last chance. It was the opinion of the treating clinicians that the alternative was palliation. Given the patient was informed of this rationale and agreed it was entirely appropriate. In such circumstances methylprednisolone seems a reasonable choice, 2mg/kg loading followed by a prolonged reducing regime. Outside of such, relatively unusual, parameters I would not support its use. References 1. The ARDS Definition Task Force. Acute Respiratory Distress Syndrome: The Berlin Definition. JAMA 2012; 307(23): 2526-2533. 2. Meduri GU et al. Methylprednisolone infusion in early severe ARDS. Chest 2007; 131: 954-963. 3. The ARDS Network. Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. NEJM 2000; 342(18): 1301-1308. 4. Annane D, Sebille V & Bellissant E. Effect of low doses of corticosteroids in septic shock patients with or without early acute respiratory distress syndrome. Crit Care Med 2006; 34(1): 22-30. 5. Bernard GR et al. High-dose corticosteroids in patients with the adult respiratory distress syndrome. NEJM 1987; 317(25): 1565-1570. 6. The ARDS Network. Efficacy and safety of corticosteroids for persistent acute respiratory distress syndrome. NEJM 2006; 354(16): 1671-1684. 7. Meduri GU et al. Effect of prolonged methylprednisolone therapy in unresolving acute respiratory distress syndrome. JAMA 1998; 280(2): 159-165. 8. Ruan S et al. Exploring the heterogeneity of effects of corticosteroids on acute respiratory distress syndrome: a systematic review and meta-analysis. Critical Care 2014; 18: R63. 9. Guerin C et al. Prone positioning in severe acute respiratory distress syndrome. NEJM 2013; 368(23): 2159-2168.