Rationale for Prolonged Glucocorticoid Treatment in ARDS

Supplementary Material

Prolonged methylprednisolone treatment is associated with improved ARDS outcomes: analysis of individual patients’ data from four randomized trials and review of the literature.

Rationale for prolonged glucocorticoid treatment in ARDS

In ARDS, resolution of inflammation and restoration of tissue homeostasis is an active, continuous and coordinated process (switch from production of pro-inflammatory towards pro-resolving molecules [1]) – characterized by non-phlogistic elimination of granulocytes by apoptosis, restoration of the alveolar-capillary membrane’s integrity, resorption of edema, increased surfactant production, decreased clotting and fibroproliferation with resolution of intra-alveolar and interstitial granulation tissue [2, 3]- that can be accelerated by prolonged glucocorticoid treatment [4].There is a strong scientific foundation in support of prolonged methylprednisolone treatment in ARDS (see Supplementary Material), with evidence showing that treatment positively affects all "layers" – pathogenesis (biology - core stratum), morphology (histology – intermediate stratum) and clinical (physiology - outer stratum) – of the disease process [5].

In ARDS, inadequate GC-glucocorticoid receptor  (GR) -mediated down-regulation of pro-inflammatory transcription factor nuclear factor-B (NF-B) in circulating and tissue cells leads to persistent elevation over time (> 4 weeks) in plasma and bronchoalveloar lavage (BAL) markers of inflammation, hemostasis, and tissue repair (maladaptive repair)[3]. Experimental ARDS is associated with a significant reduction in lung tissue GRnumber and binding capacity, and glucocorticoid treatment – contrary to placebo - restores GRnumber and function leading to resolution of pulmonary inflammation [6]. Moreover, alveolar macrophages extracted from patients with established ARDS have decreased 11β-HSD oxo-reductase activity with decreased conversion of cortisone to cortisol [7].

In an ex-vivo study, we have shown that prolonged methylprednisolone treatment – contrary to placebo – was associated with upregulation in GRnumber and nuclear translocation with reduction in NF-B DNA-binding and transcription of inflammatory cytokines [8]. ARDS patients randomized to prolonged methylprednisolone treatment, contrary to placebo, demonstrated a sustained reduction in plasma and/ or BAL levels of tumor necrosis factor- (TNF-, interleukin (IL)-1, IL-6 [3], IL-8, soluble intercellular adhesion molecule-1 [3], procollagenaminoterminalpropeptide type I and III [3], indices of alveolar-capillary membrane permeability (BAL albumin, total protein, and percentage neutrophils) [9], and an increase in IL-10 [3], protein C [10], and surfactant [11].

Contrary to older trials investigating a time-limited massive daily methylprednisolone dose[12], low-to-moderate dose glucocorticoid treatment was not found – in this study and prior meta-analyses [12, 13] - to be associated with increased risk for nosocomial infections. Lowering systemic and pulmonary inflammation might lower the risk for developing nosocomial infection by (i) decreasing duration of MV and ICU stay, (ii) achieving an inflammatory milieu less favorable to intra and extra cellular growth of bacterial pathogens frequently encountered (Staphylococcus aureus, Pseudomonas aeruginosa, and Acinetobactersps.) [14], and (iii) improving opsonization-dependent phagocytic neutrophil function [15] and intracellular killing [16].

Methods

In the methylprednisolone RTCs (Table S1), protocols allowed for early discontinuation of study drug for development of complications (shock[17], disseminated fungal infection[17, 18], gastro-intestinal bleeding requiring transfusion[18, 19]) or to received glucocorticoid treatment (failure to improved lung injury score (LIS)[20] variables within 7-14 days of randomization[18, 19], or development of life-threatening conditions requiring glucocorticoid treatment [18, 19]. Per protocol, duration of glucocorticoid treatment after removal of AB (Table S2) was up to 14 days in the early ARDS trials[19, 21], and varied between 18 days [18]and 0-4 days[17]in the late ARDS trials.

Mechanical ventilation and ICU-free days to day 28 for patients who die before or on study day 28 was set to 0, and for patients surviving beyond day 28 was defined as the number of days (1-to-28) that the patient was successfully without MV or outside the ICU. Return to AB or ICU were included in the calculation of total duration of MV or ICU stay to study day 28. Timing of and reason for return to AB or ICU were available in two trials[18, 19], and not available in the ARDS network trial[17]. For return to AB before study day 28, we applied the ARDS network trial’s definition[17] (study termination form - CRF instructions version 3; 7/6/01) stating if the patient achieved 48 hours of UAB but returned to assisted breathing before day 28, record the number of calendar dates on which the patient received AB for any period of time.

We searched PubMed and Google Scholar for reports in any language published before March 1, 2015, using the search terms “acute lung injury,” “acute respiratory distress syndrome,” “acute respiratory failure,” “pneumonia,” “steroids,” “glucocorticoids,” and ”corticosteroids.” We identified four trials [17-19, 21] that investigated prolonged methylprednisolone and four trials [22-25] that investigated prolonged hydrocortisone treatment in ARDS. The updated literature search identified four randomized trials investigating prolonged hydrocortisone treatment in patients with early ARDS [22-24]. The largest study[23] is a post hoc analysis of 177 patients with ARDS from a trial of vasopressor-dependent septic shock [26]. In this trial [26], the subgroup (n=129) of non-responders to a short corticotropin test, contrary to responders (n=48) had – in response to hydrocortisone treatment - more days alive without AB (5.6 ± 8.8 vs. 2.6 ± 6.6; p =0.006) and improved survival (47% vs. 25%; p =0.021). One small trial (Chines language) investigated patients with early ARDS and critical illness corticosteroid insufficiency [25]. This trial reported a non-significant reduction in ICU mortality 17% vs. 50% (p>0.05) [25]. The other two trials[22, 24] of patients with severe community-acquired pneumonia included 94 patients on mechanical ventilation with early ARDS. In both trials [22, 24], hydrocortisone treatment was associated with significant reduction in duration of AB; one trial reported improved survival [22]. With the exception of hyperglycemia, glucocorticoid treatment was not associated with increased rate of side effects.

Results

Patients’ baseline characteristics of the four trials investigating prolonged methylprednisolone treatment in ARDS are shown in Table S1.Reason for discontinuing study drug before achieving UAB was available in 14 patients in the ARDS network trial[17] (11 shock, 3fungal infections) and in the seven patients of the early ARDS trial[19] (failure to improve with cross-over to open label methylprednisolone (2), HIV - exclusion criteria - diagnosed after study entry, withdrawal of care (2), family request, and vasculitis).

The methylprednisolone-treated group had a significantly shorter duration (days) of initial AB by day 28 (Table S2). The ARDS network trial was the only trial with more patients randomized to active drug that returned to assisted breathing after rapid tapering. In the ARDS network trial (N = 20),[17] reasons for return to AB were not provided. In the other ARDS trials (N = 8)[18, 19], none of the patients that returned to AB had recrudescence of ARDS; reasons for return to AB included: 3 severe sepsis (1 septic shock), 2 difficulty-clearing secretions (1 placebo), 1 cardiogenic pulmonary edema, 1 post-operative observation, and 1 pleural effusion with lung trapping[18, 19].

The methylprednisolone-treated group also had significantly more patients discharged alive from the ICU by day 28 (Table S3). The ARDS network trial[17] was the only trial with more patients randomized to active drug that returned to the ICU. For the ARDS network trial[17], reasons for return to ICU were not available; ten of 18 patients that retuned to ICU also returned to AB.[17] In the other three trials[18, 19, 21], seven patients returned to ICU (one placebo), 5 for return to AB (see above), 1 for postoperative observation, and 1 for cardiac event; 2 died (active drug). Tables S4show the Cox Proportional Hazard Models for all patients and patients randomized before day 14 of ARDS onset with Competing Risk adjustment for time to hospital death by Day-28.

For the trial-level meta-analysis, we evaluated risk of bias for each important outcome (Table S5). Subgroup analyses are described in Table S6. RCTs enrolling <60 patients had consistently larger effect estimates across all beneficial outcomes. The one single-blind RCT and the one trial published after 2010 had a larger effect estimate, but these subgroup findings are less credible because they were not consistent across outcomes and the p-values are relatively large (p=0.02 for both). There were no differences for any other subgroup analysis. In a sensitivity analysis excluding RCTs with a sequential stopping rule [18, 19, 22],the benefits of glucocorticoids did not retain statistical significance for ICU-free days or mortality. This was also true for mortality when the sensitivity analysis was restricted to RCTs with ≥60 patients (mortality RR 0.83, 95% CI 0.69 to 1.01).Mechanical ventilation-free day was robust to exclusion of studies with sequential stopping rules.

TABLES

FIGURES

Figure S1. Patients, in each trial, alive on study day 8, 15, 22, and 29 that received at least at least 7, 14, 21, and 28 days of methylprednisolone treatment.

Legend: N value within each column reports patients alive on the specified day.

* = P < 0.001 in comparison to the other three trials

Figure S2.Probability of achieving (initial) unassisted breathing (methylprednisolone vs. placebo) from randomization to day 28 in trials with early vs. late initiation of treatment.

Legend: Cumulative incidenceestimates - adjusted for competing risk - of the probability of achieving (initial) unassisted breathing by day 28 for patients with acute respiratory distress syndrome in trials with early ( 3 days) vs. late (5 days) initiation of treatment.

Figure S3. Effects of prolonged glucocorticoid treatment on mechanical ventilation-free days (n= 510) and intensive care unit-free days (n= 354) at day 28.