Grand Rounds--Hammersmith Hospital: Thromboembolic occlusion of a pulmonary arteriovenous malformationBMJ 1995; 311 doi: https://doi.org/10.1136/bmj.311.7004.553 (Published 26 August 1995) Cite this as: BMJ 1995;311:553
Case presented by: Ian Sabroe, registrar in respiratory medicine. Chairman: S Bloom, professor of endocrinology. Discussion group: J M B Hughes, professor of respiratory medicine S Shaunak, senior lecturer in infectious diseases I Robertson, senior registrar in radiology. Series edited by: Dr W A Lynn.
Pulmonary embolism is a common medical emergency, typically associated with hypoxia secondary to a ventilation perfusion mismatch. Pulmonary arteriovenous malformations are rare and cause hypoxia due to direct arterial venous shunting. We present the case of a woman with both pulmonary embolism and a pulmonary arteriovenous malformation in whom an embolus fortuitously occluded the malformation leading to a paradoxical improvement in oxygenation.
A 74 year old woman was admitted to her local hospital with a history of rapidly worsening shortness of breath over the preceding few weeks. She had been suffering from slowly progressive dyspnoea for one to two years, and over the past six months her friends had noticed that her lips and fingers were bluish. Recently she had also become short of breath at rest. She had smoked 40 cigarettes a day for 55-60 years, and chronic bronchitis had been diagnosed several years before. Five years ago, after a minor episode of haemoptysis, plain chest radiography had shown a probable pulmonary arteriovenous malformation. She refused further investigation or treatment of this. She had no history of nose bleeds or overt gastrointestinal bleeding and no neurological symptoms.
On admission to her local hospital she was hypoxaemic and mildly hypercapnic on air (arterial oxygen pressure 6.6 kPa and alveolar carbon dioxide pressure 6.1 kPa). The hypoxaemia was unresponsive to oxygen therapy, suggesting the presence of a right to left shunt. Chest radiography again confirmed the presence of a possible pulmonary arteriovenous malformation. Four weeks after admission she was transferred to this hospital.
On admission she was cyanosed, with a respiratory rate of 30 breaths/min, pulse of 100 beats/min, and blood pressure of 130/80 mm Hg. She had no telangiectases and no family history of hereditary haemorrhagic telangiectasia. The lungs were hyperinflated with occasional wheeze. She had no cor pulmonale and no cardiac murmurs or pulmonary bruits. Full blood count was normal (haemoglobin concentration 15.5 g/l). Pulmonary function tests showed a reduced vital capacity (65% of predicted) and a ratio of forced expiratory volume in one second to vital capacity of 0.50, consistent with chronic airflow obstruction. The electrocardiogram showed small inferior Q waves, and the chest x ray film showed an apparent pleural shadow in the left mid-zone but no obvious pulmonary arteriovenous malformation. Measurements of arterial blood gas pressures on air and oxygen showed a normal alveolar carbon dioxide pressure but an arterial oxygen pressure on air of 8.1 kPa rising to 14.6 kPa on 35% oxygen --consistent with ventilation perfusion mismatch rather than shunting. To look for the presence of pulmonary shunts, a lung perfusion scan was performed with technetium labelled macroaggregates of albumin with additional (gamma) camera images over the right kidney. When intrapulmonary shunting is present, labelled aggregates may pass through to the systemic circulation, and the size of the shunt is then calculated from the amount of radioactivity detected in the right kidney compared with the amount injected. The scan in this patient showed no intrapulmonary shunt but showed multiple perfusion defects highly suggestive of pulmonary emboli. A ventilation scan confirmed that the defects were not matched. No source of thromboembolism was found despite investigations including Doppler ultrasound of the leg and iliac veins.
The patient was treated with heparin then warfarin, with an initial improvement in symptoms. Eight days later she became more breathless and cyanosed. Her arterial oxygen pressure was now 7.2 kPa on air with no response to oxygen therapy of </=60% fractional inspired oxygen. A ventilation perfusion scan showed almost full resolution of the pulmonary emboli but accumulation of the technetium labelled albumin aggregates in the kidney, and chest radiography now showed a clear, left mid-zone pulmonary arteriovenous malformation. Pulmonary angiography showed a large, simple, solitary pulmonary arteriovenous malformation and normal pulmonary arterial pressures. The size of the shunt was calculated to be 35% of resting cardiac output.
Over the next few days she became more hypoxaemic and breathless and was therefore considered for embolisation. As she had been treated with anticoagulant drugs, balloon occlusion of the feeding vessel was attempted, but the local anatomy proved unfavourable and occlusion of the pulmonary arteriovenous malformation was not possible. Anticoagulation was therefore briefly discontinued to allow occlusion by placement of fibrin coated steel coils, eight of which were needed to block completely the shunt as assessed by angiography during the procedure. After embolisation her arterial oxygen pressure immediately rose to 9 kPa on air with a further brisk increase on supplemental oxygen. Technetium lung scanning confirmed occlusion of the pulmonary arteriovenous malformation and resolution of the pulmonary emboli. Her exercise tolerance improved appreciably, and she was discharged well taking warfarin.
The chronic hypoxaemia in this patient was due to a combination of a large pulmonary arteriovenous malformation and moderately severe emphysema. During admission to her local hospital she developed pulmonary emboli--presumably secondary to immobilisation on the ward--which occluded the pulmonary arteriovenous malformation. The malformation became patent again after she was treated with anticoagulant drugs, leading to a return of her hypoxaemia and breathlessness. The lung normally acts as a filter for thrombi, preventing their progression into the systemic circulation where they may cause serious sequelae such as stroke, so she was extremely lucky not to develop potentially fatal paradoxical emboli. The occurrence of both pulmonary emboli and pulmonary arteriovenous malformations is extremely rare, and only one previously documented case exists.1
Eighty per cent of pulmonary arteriovenous malformations are associated with hereditary haemorrhagic telangiectasia and are usually multiple. Pulmonary arteriovenous malformations not associated with hereditary haemorrhagic telangiectasia (such as in our patient) are usually single. The malformations are often visible on chest x ray films, and an overlying bruit may be heard. The size of a shunt used to be calculated from the differences in expected and observed arterial oxygen concentrations when the patient is breathing 100% oxygen. This has been superseded by a radioisotope method developed at this hospital, which has proved to be accurate; the size of a shunt can also be measured on exercise.2 3
The complications of pulmonary arteriovenous malformations are predominantly chronic hypoxaemia (secondary to the right to left shunting through the malformation) and paradoxical embolism. In the two major series of patients with hereditary haemorrhagic telangiectasia at this hospital4 and Johns Hopkins Hospital,5 cerebral abscesses occurred in approximately 20% of patients, stroke in 18%, and transient ischaemic episodes in 37%. In the Johns Hopkins series 59 patients had routine computed tomography of the head, which showed evidence of previous stroke in 36% of scans.
Pulmonary arteriovenous malformations were first treated surgically in 1942, but more recently surgery has been superseded by angiographic embolisation.4 The method of choice for the occlusion of these malformations in Britain is the placement of fibrin coated steel coils in the feeding artery of the malformation, which occlude it by thrombosis at that site. These coils may not be effective in a patient who has been treated with anticoagulant drugs, particularly heparin. The alternative procedure of balloon occlusion of the feeding vessel is widely practised in the United States and relies less on effective thrombosis, making it more suitable when anticoagulation is needed. Balloon occlusion proved impossible in the case of our patient, and therefore coil occlusion was done during a brief interval when she was not receiving anticoagulant drugs.
SB: Is there any possibility of the shunt reopening?
JMBH: We have performed embolisation in 64 patients so far, none of whom have reopenedshunts that have been occluded. Patients with hereditary haemorrhagic telangiectasia who become pregnant commonly develop new shunts (probably owing to the enlargement of existing small shunts); we have seen this on several occasions. This was an extraordinary case, and the patient was extremely lucky not to develop serious paradoxical embolism. It is interesting to speculate that the shunt was probably very small in the past, but, as her emphysema worsened and her pulmonary arterial pressure rose, the size of the shunt increased, causing the cyanosis and dyspnoea on exertion. Normally, we prefer to embolise these patients with coils, as balloons cannot be placed so precisely in the vessel which feeds the pulmonary arteriovenous malformation; proximal placement may cause substantial loss of normal lung tissue. In this case the balloon procedure was tried while the patient received anticoagulant drugs, but the vascular anatomy was unfavourable. It may be better for her that she ended up with a more selective embolisation with fibrin coated coils.
SB: Had there been any previous pulmonary embolisation that might also have increased the size of her shunt?
JMBH: No. Her right heart pressures were normal an angiography, although the shunt itself acts as an escape valve for increased pulmonary arterial pressures.
SS: How long do you follow up patients for evidence of shunt recanalisation?
JMBH: All patients receive lung scans three months after embolisation to confirm shunt obliteration. Long term follow up is done with oximetry as this has proved to be a very reliable screening test for shunts. Patients with both hereditary haemorrhagic telangiectasia and pulmonary arteriovenous malformations show a drop in arterial oxygen pressure when standing, as most pulmonary arteriovenous malformations are distributed basally in the lungs and are more selectively perfused in the upright position.
SB: Does thrombosis extend beyond the steel coils? How important is the fibrin coating of the coils?
JMBH: Thrombosis occurs only on the coils. Radiolabelled platelet scans performed at the time of occlusion show that this thrombosis occurs within minutes.
IR: Stainless steel by itself is poorly thrombogenic in this setting. The fibrin coated coils work by first distorting the intima of the blood vessel in which they are placed and causing the release of locally active mediators which cause platelet aggregation; this is then enhanced by the fibrin coating, which allows rapid and permanent thrombogenesis.
The BMJ welcomes grand rounds from other hospitals.