Pleural effusions in patients with chronic myeloid leukaemia treated with dasatinib may have an immune-mediated pathogenesis


The development of pleural effusions is a frequent complication of dasatinib therapy (Cortes et al, 2007; Guilhot et al, 2007; Hochhaus et al, 2007); it has been reported in 14–35% of chronic myeloid leukaemia (CML) patients on dasatinib, and this may necessitate interruptions and/or reductions in the dose of dasatinib (Cortes et al, 2007; Guilhot et al, 2007; Hochhaus et al, 2007). Several risk factors for the development of pleural effusions have been identified, but the aetiology of the effusion remains unclear.

We retrospectively analysed our series of 62 consecutive CML patients treated with dasatinib at Hammersmith Hospital to identify factors predictive for the development of pleural effusions. Chest X-rays were performed in all patients with respiratory symptoms. Clinical management included temporary or permanent discontinuation of the dasatinib, dose modification, administration of steroids and/or diuretics and thoracocentesis. The probability of pleural effusion was calculated using the cumulative incidence procedure. Univariate analysis to identify prognostic factors (Table I) for the development of pleural effusion was carried out using the log-rank test. Variables found to be significant at the P < 0·25 level were entered into a proportional hazards regression analysis. The influence of the development of skin rash during dasatinib therapy on the development of pleural effusion was studied using a time dependent Cox model. P-values were two-sided, with values <0·05 considered significant, and confidence intervals (CI) refer to 95% boundaries.

Table I.   Univariate analysis of variables associated with an increased risk of pleural effusion.
VariablesNumber of patientsCumulative incidence of pleural effusion at 1 year (%) P-value
  1. Cardiac history was defined as coronary artery disease, atrial fibrillation or congestive heart failure; history of lung disease included chronic obstructive pulmonary disease, asthma, pneumonia, and pulmonary fibrosis. Other variables were explored and found not to be statistically significant, namely: duration of disease, prior interferon, duration of imatinib, starting dose of dasatinib.

  2. CML, chronic myeloid leukaemia; CP, chronic phase; AP, acute phase; BP, blastic phase.

CML phase of the disease  0·014
 CP + AP5340·1 
Age  0·066
 >55 years3840 
 <55 years24 9 
Cardiac history  0·017
Hypertension  0·033
Hypercholesterolemia  0·019
History of lung disease  0·31
Fluid retention  0·94
History of auto-immune disease  0·0008
History of skin rash on imatinib  0·004
 Grade 1–41270·9 

Median age was 58·2 years (range, 29–79 years). Among them 46 (74%) were in chronic phase (CP), seven (11%) in accelerated phase and nine (15%) in blastic phase. Forty-six patients started dasatinib at a dose of 70 mg twice daily, nine at 50 mg twice daily, three at 140 mg once daily and four at 100 mg once daily. Seventeen patients (27%) developed respiratory symptoms and pleural effusions at a median of 179 d from the initiation of dasatinib therapy (range, 20–756 d). The cumulative incidence of pleural effusion at 1 year was 29·5% (95% CI 18·6–43·3%). Among the patients who developed effusions, 14 were of grade 2 and 3 were of grade 3 severity. Fifteen were unilateral, and two bilateral; two had concomitant grade 2 pericardial effusions on echocardiography without any sign of pulmonary hypertension. As described by others (Bergeron et al, 2007), additional lung abnormalities were found in four patients, namely left lower lobe consolidation (n = 2), pulmonary nodular opacities and septal thickening (n = 1), and alveolar shadowing (n = 1). The actual dasatinib dose at the time of detection of the pleural effusion was 140 mg per day in 11 (65%) of the 17 patients.

We investigated the impact of several variables on the development of pleural effusion (Table I). Blastic phase CML, previous history of cardiac disease, hypertension and hypercholesterolemia were significantly associated with an increased risk of pleural effusion. Our findings are consistent with a recent published study (Quintas-Cardama et al, 2007), where a higher incidence of pleural effusions was seen in patients with hypertension, a history of cardiac disease and a dasatinib dosage >100 mg per day. Because an immune-mediated mechanism has been suggested as a possible explanation for pleural effusion under dasatinib (Bergeron et al, 2007), we investigated different immune-mediated events among those patients. Interestingly eight of the 62 patients had a previous history of auto-immune disease (hypothyroidism n = 4, Sweets syndrome n = 1, Stevens-Johnson syndrome n = 1, hyperthyroidism n = 1, Sjögren Syndrome-arthritis n = 1). Notably, two patients had developed hypothyroidism while on interferon alpha. A prior history of auto-immune disease was strongly associated with the risk of developing a pleural effusion (see Table I). Moreover, a prior history of skin rash on imatinib was also associated with an increased risk of pleural effusion so we queried the significance of skin rashes developing on dasatinib. Fourteen patients experienced a skin rash while taking dasatinib (12 patients with grade 2–4 skin rash, of whom three also had panniculitis); eight of these patients developed a pleural effusion, which occurred in each case after the onset of the rash. Thus the occurrence of a rash attributable to dasatinib was significantly associated with the risk of pleural effusion (RR = 7·07, 95% CI 2·6–19·2, P < 0·0001). By multivariate analysis, there were three independent predictors for the development of pleural effusion, namely a prior skin rash on dasatinib (RR = 5·3, 95% CI 1·9–14·6, P = 0·001), a previous history of auto-immune disease (RR = 4·3, 95% CI 1·3–14·0, P = 0·015) and hypercholesterolemia (RR = 3·5, 95% CI 1·8–14·6, P = 0·037).

The management of the pleural effusions included stopping the dasatinib and treating all patients with diuretics. Three patients had a thoracocentesis, one received steroids and one had a pleurodesis. Among the patients who had a pleural effusion, seven (41%) were still on dasatinib at the last follow-up and 10 (59%) had discontinued dasatinib due to disease progression (n = 6), recurrent pleural effusion (n = 3) or to undergo stem cell transplantation (n = 1).

The mechanism by which dasatinib induces a pleural effusion associated with dasatinib is unclear. It has been suggested that inhibition of PDGFRB, which is expressed in pericytes and is involved in the regulation of angiogenesis, may be implicated (Jayson et al, 2005; Quintas-Cardama et al, 2007). However other mechanisms have been suggested. Bergeron et al (2002, 2007) suggested a possible immune-mediated mechanism based on the high lymphocyte frequency in pleural fluids and pleural tissue in those patients, the presence of auto-antibodies reported in one patient and the response to steroids. Our results support a possible immune origin, as the effusions seemed to be associated with other immune-mediated reactions, such as skin rash on imatinib or dasatinib or prior history of auto-immunity. This association might be reinforced by the fact that TEC kinase and Bruton agammaglobulinemia tyrosine kinase (BTK), both targets for dasatinib, are involved in the signalling pathway of the T- and B-cell receptors as well as in endothelial, pulmonary and mast cells (Rix et al, 2007). Thus, pleural effusion is likely to be multi-factorial where auto-immunity, inhibition of PDGFRB and cardiac dysfunction all play the major roles. Further investigation is therefore required to address the underlying mechanism.

Author contributions

Hugues de Lavallade: collected and analysed clinical data, and wrote the manuscript. Sinthiya Punnialingam: collected and analysed clinical data. Dragana Milojkovic: provided clinical care and commented on the manuscript. Marco Bua: provided clinical care and commented on the manuscript. Jamshid S. Khorashad: collected and analysed clinical data, and commented on the manuscript. Ian H. Gabriel: provided clinical care and commented on the manuscript. Aristeidis Chaidos: provided clinical care and commented on the manuscript. Eduardo Olavarria: provided clinical care and commented on the manuscript. John M. Goldman: commented and revised the manuscript. Jane F. Apperley: responsible for coordinating the CML program, commented and revised the manuscript. David Marin: designed the study, supervised patient care, performed the statistical analysis, and wrote the manuscript.


We are grateful for support from the NIHR Biomedical Research Centre Funding Scheme; Hugues de Lavallade was supported by a grant from the ‘Fondation de France’ (Paris, France).