A 64-year-old woman with unremarkable medical history visited her general practitioner in October 2011 with facial swelling. Specific exposures provoking these symptoms could not be identified, and after several days the swelling disappeared spontaneously. In November she presented at an affiliated hospital with progressive dyspnea, nonproductive cough, and fatigue, without fever or orthopnea. On examination the patient appeared comfortable, without signs of respiratory distress. The blood pressure was 110/54 mm Hg (baseline measurements: 142/90 mm Hg in 2010 and 140/80 mm Hg in 2011), the heart rate 67 beats/minute, temperature 36.9°C, and oxygen saturation 97% at ambient air (Fi_O2 21%). She had a normal jugular venous pressure, but bibasilar rales in the lungs, periorbital edema, and pitting edema in both legs. She reported a recent increase in bodyweight of 10 kg. Blood tests indicated hemoconcentration (hemoglobin 9.6 mM or 15.5 g/dl) and hypoalbuminemia (albumin 30 g/L). Blood count, electrolytes, and inflammatory markers were normal, as were renal function and liver enzymes. Chest radiography showed bilateral pulmonary edema, pleural effusion, and atelectasis. Cardiac failure was excluded by normal ventricular function on cardiac echography and normal N-terminal pro–brain natriuretic peptide levels (<121 ng/L). Other noninfectious causes of general edema, like anaphylaxis, C1-esterase inhibitor deficiency, malignancy (by positron emission tomography and computed tomography imaging), nephrotic syndrome, protein losing enteropathy, adrenal insufficiency, and systemic mastocytosis, were excluded. Because of steadily increasing dyspnea, our patient was treated with diuretics and underwent several thoracenteses. On further deterioration of clinical condition, she was referred to our academic center for analysis (March 2012).

Systemic capillary leak syndrome was considered, because serum analysis revealed the presence of M-protein (IgGκ 0.21 g/dl), a protein associated with capillary leak syndrome (3). However, the chronic character of the vascular leak and the presence of pleural effusions rendered systemic capillary leak syndrome less likely in our patient. Putative markers of systemic capillary leak syndrome, vascular endothelial growth factor and angiopoietin-2 (1, 4), were not detectable (both <0.1 ng/mL). To evaluate the severity of the vascular leak in the lungs we measured the pulmonary leak index, which determines the permeability of the lung vasculature on the basis of distribution of radiolabeled protein (5). The pulmonary leak index in our patient was 23 × 10⁻³/min, revealing substantially increased permeability of the lung vasculature (normal <14 × 10⁻³/min), and approaching values found in patients with acute respiratory distress syndrome (>30 × 10⁻³/min) (5). On the basis of these measurements, the increased body weight, mild hypotension, hypoalbuminemia, the hemoconcentration, and the absence of a clear eliciting condition, idiopathic vascular leak was diagnosed.

Because imatinib (Gleevec, Basel, Switzerland) was recently shown to protect against vascular leak (6), we started treatment with a low dose imatinib (200 mg/d). One month after initiation of imatinib treatment, the dyspnea had largely disappeared, which was demonstrated by a reduced requirement for thoracentesis. Simultaneously, body weight normalized and plasma albumin levels increased (Figure 1A), together with disappearance of peripheral edema. Chest radiography demonstrated reduced amounts of pleural fluid (Figure 1B). In August, the pulmonary leak index was 17 × 10⁻³/min, indicating reversal of the increased pulmonary vascular permeability (Figure 1A). The dyspnea had completely disappeared. No side effects of imatinib were observed. The patient consented to off-label use of imatinib, and provided written informed consent for publication of the case report. The institutional medical ethical committee gave permission for off-label use of imatinib, with reference to compassionate use in the acute setting.

Figure 1. Parameters of vascular leak before and after initiation of imatinib treatment. (A) Time curves of body weight (blue) and albumin levels (red) as indirect parameters of vascular leak, and the pulmonary leak index (green) as direct parameter of vascular leak. The average pulmonary leak index of both lungs is given together with the range of the values measured for the two lungs (error bars). The gray area indicates the period of imatinib treatment (200 mg/d). Black triangles represent thoracenteses, of which volumes ranged from 50 ml (diagnostic thoracenteses) to 1,500 ml (therapeutic thoracenteses). (B) Chest radiographic examinations for evaluation of pleural fluid. The image on the left reveals the presence of bilateral pleural fluid, whereas the image on the right shows clear contours of the heart and the diaphragm, suggesting that pleural fluid substantially improved after imatinib.

Download Figure | Download Powerpoint

[More] [Minimize]

To the best of our knowledge, this is the first description of treatment of vascular leak with imatinib in a controlled setting. Generalized loss of endothelial barrier integrity forms a key pathophysiological mechanism in (idiopathic) vascular leak. Soluble factors induce contraction of endothelial cells and subsequent gap formation in the endothelial monolayer (1). The vascular leak that follows disruption of the endothelial barrier carries high mortality in a wide variety of diseases. Laboratory studies recently showed that imatinib protects against endothelial barrier disruption and vascular leak by strengthening endothelial cell–cell contacts and adhesion of endothelial cells to the extracellular matrix (6). Experimental treatment of our patient with imatinib was followed by normalization of direct (pulmonary leak index and necessity of thoracentesis) and indirect (body weight, albumin levels, hemoglobin levels, and dyspnea) parameters of vascular leak.

The observation that imatinib protects against vascular leakage is surprising in light of previous studies that report peripheral edema as a side effect of imatinib (7). This paradoxical finding may first of all be explained by duration and intensity of imatinib treatment. Compared with patients treated with imatinib for chronic myeloid leukemia, imatinib treatment in our patient was relatively short, and the imatinib dose given to our patient was low (200 mg/d). In addition, the mechanism of edema formation in idiopathic vascular leak likely differs from the mechanism of edema formation as a side effect of imatinib treatment. In idiopathic vascular leak, onset of edema is (sub-)acute, and follows dysfunction of the endothelial barrier (by either contraction or apoptosis of endothelial cells) (1). Although the mechanism of edema formation during imatinib treatment remains incompletely understood, it has been suggested that chronic inhibition (several months to years) of growth factors like platelet-derived growth factor receptor impairs pericyte function (8), leading to weakening of the vessel wall. This may explain the more gradual and chronic onset of edema as observed during long-term imatinib treatment. The use of imatinib in the therapeutic manner employed on our patient may have important implications for conditions that share endothelial barrier dysfunction as a common pathophysiological mechanism. The application of imatinib for reversal of endothelial barrier dysfunction may therefore include various conditions like systemic capillary leak, sepsis, and acute respiratory distress syndrome, which currently lack specific therapy (1–3).

Causal treatment for vascular leak is currently lacking. Although phase I/II trials are required to assess safety of imatinib in vascular leak–associated conditions, the current case suggests that imatinib may seal the gap.