Dexamethasone in patients with acute lung injury from acute monocytic leukaemia

RESULTS

From January 2005 to December 2008, 450 patients with newly diagnosed acute myeloid leukaemia were admitted to the Saint-Louis Teaching Hospital, including 45 (10%) with AML FAB-M5, of whom 20 (44%) required ICU admission before chemotherapy initiation for respiratory manifestations with onset before the diagnosis of leukaemia (fig. 1). Of these 20 patients, 11 were admitted directly to the ICU for acute respiratory failure and nine were transferred from the haematological wards within 12 (6–36) h of admission. In 14 patients, AML FAB-M5 was diagnosed during the ICU stay.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1–

Patient flow chart. ICU: intensive care unit; AML5: acute myelocytic leukaemia FAB-M5. ^#: the study periods ran from January 1994 to July 2002 and from January 2005 to December 2008.

As reported in table 1, there were 13 males and seven females with a median age of 42 (33–60) yrs. None had any comorbidity. All patients had severe acute respiratory failure with tachypnoea and profound hypoxaemia. The chest radiographs consistently showed diffuse lung infiltrates. The Simplified Acute Physiology Score version II score was 39 (29–50). All 20 patients were febrile and all but one had physical evidence of leukaemic masses. Laboratory findings included high white blood cell count in 17 patients, circulating blast cells in 16 and platelet count <50 1×10⁹ cells·L⁻¹ in 12. The three patients without hyperleukocytosis had <10 1×10⁹ cells·L⁻¹ with no circulating blast cells. Disseminated intravascular coagulation was present in seven patients.

As shown in table 2, within a few hours after chemotherapy initiation, 10 patients had no respiratory status deterioration or increase in oxygen requirements and 10 had a deterioration in oxygenation requiring invasive mechanical ventilation. An additional patient required intubation and ventilation because of complex ventricular arrhythmia. Median length of ICU stay was 7 days (3.5–14). ICU and hospital survival rates were both 80% (four deaths). All 16 survivors were in remission after neutropenia recovery and received consolidation chemotherapy, and 1-yr survival was 60% (eight deaths).

None of the baseline characteristics differed significantly between the historical controls and the dexamethasone-treated patients (tables 1 and 2). In particular, the severity of respiratory failure and characteristics of the leukaemia were well balanced between the two groups.

In-ICU mortality was significantly lower in the dexamethasone group than in the control group (20% versus 50%; p=0.04). In the dexamethasone-treated group, we found decreases in the occurrence of respiratory status deterioration, need for increased oxygen flow and need for ventilatory support compared with the controls (fig. 2). Chemotherapy initiation was followed by respiratory status deterioration in only four (20%) dexamethasone-treated patients compared with 20 (100%) controls.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2–

Impact of dexamethasone therapy on clinical outcomes. An additional patient from the dexamethasone group received ventilation for complex arrhythmia. ▒: dexamethasone-treated patients; ▪: historical controls. ICU: intensive care unit. *: p<0.05; **: p<0.01.

Dexamethasone therapy was not associated with increased rates of hospital-acquired bacterial or invasive fungal infections.

Table 3 reports the results of the multivariate analysis for factors associated with ICU mortality. Higher respiratory rate and lower arterial oxygen saturation measured by pulse oximetry at ICU admission were independently associated with ICU mortality. Dexamethasone therapy was associated with a trend for a decrease in ICU mortality by multivariate analysis.

DISCUSSION

We evaluated the efficacy and safety of dexamethasone therapy in patients with AML FAB-M5 and noninfectious pulmonary infiltrates. Compared with historical controls who did not receive dexamethasone, patients given dexamethasone had a decrease in the severity of pulmonary involvement and a lower rate of respiratory status deterioration after chemotherapy initiation. ICU mortality was lower in the dexamethasone-treated patients than in the historical controls.

Pulmonary leukostasis occurs in patients with acute myeloid leukaemia and rapidly increasing white blood cell count [10, 11], and is consistently present when the count exceeds 200 1×10⁹ cells·L⁻¹ [12]. Pulmonary leukostasis leads to endothelial injury and activation from microvascular invasion, with hyperviscosity, leukocytic microthrombi, oxygen steal and hypoxia [13–15]. Leukaemic pulmonary infiltrates occur in patients with or without hyperleukocytosis. This fact suggests that both the type of the blasts and their affinity for the pulmonary endothelium may be responsible for lung injury [8, 28]. Autopsy studies have shown blast aggregates within the vessel lumina [29]. The infiltrates typically follow the lymphatic routes along the bronchovascular bundles, interlobular septa and pleural interstitial tissue [16–18]. Finally, lysis pneumopathy occurs either immediately or early after chemotherapy initiation as a manifestation of acute tumour lysis syndrome. Lysis pneumopathy results in diffuse alveolar damage [30] and is most common in patients with hyperleukocytic acute myeloid leukaemia, particularly the myelomonocytic subtypes with abnormal marrow eosinophils [31].

Steroids have been used in early severe ARDS to improve oxygenation by decreasing lung collagen and oedema formation [32, 33]. However, steroid use in patients with persistent ARDS is controversial [21, 22]. Steroids are a major component of the treatment regimen for acute lymphocytic leukaemia but are not used in patients with acute myeloid leukaemia [34]. In patients with promyelocytic leukaemia given ATRA, dexamethasone was very effective in preventing or treating ATRA-related pulmonary toxicity [23, 35].

Noninfectious pulmonary involvement is particularly common in patients with hyperleukocytotic myeloid leukaemia, particularly of the M4 and M5 subtypes [18]. In a previous study, we described 20 AML FAB-M5 patients with acute respiratory failure from pulmonary leukostasis and leukaemic infiltration before the diagnosis of leukaemia [8]. All 20 patients had post-chemotherapy lysis pneumopathy. 10 patients died, indicating a need for an intervention targeting the pathophysiological mechanisms responsible for the initial lung injury and subsequent respiratory status deterioration. Steroid therapy was a good candidate, as steroids are widely used in various subsets of ALI [33] and chemotherapy-related pulmonary toxicity [23]. Our data suggest that steroid therapy may not only limit pulmonary leukostasis and leukaemic infiltration, but also prevent lysis pneumopathy. Compared with the control group, the dexamethasone-treated group was characterised by lower rates of respiratory status deterioration, mechanical ventilation and death. We found no significant increase in infections in the dexamethasone-treated group compared with the control group.

Our study has several limitations. We used historical controls, and the recruitment period from the first control to the last dexamethasone-treated patient spans 15 yrs, during which time changes in the management of ALI/ARDS and increasing experience may have affected patient outcomes [36]. However, the dexamethasone and control patients were not different at baseline and received the same haematological and ICU management. Our data suggest that a multicentre randomised controlled trial testing the risk/benefit ratio of dexamethasone in patients with AML FAB-M5 and noninfectious pulmonary involvement may be warranted. Whether such a trial should be extended to all patients with acute leukaemia-related noninfectious pulmonary involvement requires discussion. A second limitation of the study is that we did not record noninfectious adverse effects of dexamethasone therapy (e.g. poor glucose control). Future studies will need to record all possible adverse effects. Last, we assumed that all patients with AML FAB-M5, acute respiratory failure, pulmonary infiltrates and negative tests for infection had leukaemia-related pulmonary involvement. This reflects our standard diagnostic strategy, which is supported by five arguments. First, post mortem studies have established that the noninfectious pulmonary complications of acute leukaemia include leukostasis, leukaemic infiltration and lysis pneumopathy [8], as well as alveolar proteinosis, which is extremely rare [37]. Secondly, opportunistic infections are not encountered at the earliest phase of AML [3]. In contrast to patients with acute lymphoid leukaemia, patients with AML FAB-M5 and infection usually have community-acquired bacteria. All our patients received combination antibiotic therapy for 7 days starting at ICU admission. However, we cannot strictly rule out the diagnosis of infection, based on our data including post mortem biopsies in only two historical controls. Along this line, Confalonieri et al. [38] reported benefits from steroids in patients with severe community-acquired pneumonia. Thirdly, our study was done in a homogeneous group of patients with AML FAB-M5 patients (monocytic subtype) and inaugural respiratory failure. Among the 20 historical controls, diffuse haemorrhage by BAL and post mortem biopsies was a major finding. We believe this finding is sufficiently suggestive to maintain a high level of suspicion for leukaemia-related pulmonary infiltrates in patients with untreated AML FAB-M5 and inaugural acute respiratory failure. However, overlap may occur between lysis pneumopathy and cytarabine-induced pulmonary toxicity. Such overlap would further support steroid therapy [39]. Fourthly, all patients underwent noninvasive diagnostic tests to rule out infection [25, 27]. Also, all intubated patients underwent bronchoscopy and BAL, which consistently showed diffuse alveolar haemorrhage. Fifthly, the decrease in ICU mortality among dexamethasone patients in our study supports our presumptive diagnosis of leukaemia-related pulmonary involvement without infection. Lastly, changes in mortality between historical controls and steroid-treated cases could have been ascribable to differences in ICU management and ventilatory strategies. However, median tidal volume was 9 (6–10) mL·kg⁻¹ in historical controls and 9 (7–10) mL·kg⁻¹ in steroid-treated cases (p-value not significant). Corresponding figures for positive end-expiratory pressure were 7 (2–11) and 9 (5–15) cmH₂O (p-value not significant).

In summary, adding dexamethasone to the chemotherapy regimen in AML FAB-M5 patients with acute respiratory failure from leukaemia-related pulmonary involvement significantly diminished ICU mortality. In addition, the rate of post-chemotherapy deterioration and the need for ventilatory support decreased with dexamethasone therapy. These results suggest that dexamethasone may be effective in decreasing leukaemic pulmonary infiltration and leukostasis and in preventing lysis pneumopathy. We found no increase in infection rates with dexamethasone therapy. Although these data are not sufficient to make a recommendation about using dexamethasone, they warrant a trial of dexamethasone in patients with AML FAB-M5 presenting as acute respiratory failure without evidence of infection.