.. CHEST Original Research

Transcription

1 .. CHEST Original Research Diffuse Pulmonary Arteriovenous Malformations in Hereditary Hemorrhagic Telangiectasia* PULMONARY ARTERIOVENOUS MALFORMATIONS Long-term Results of Embolization According to the Extent of Lung Involvement Pascal Lacombe, MD; Christine Lagrange, MD; Alain Beauchet, MD; Mostafa El Hajjam, MD; Thierry Chinet, MD; and Jean-Pierre Pelage, MD, PhD Objectives: To review the safety ofembolization in patients affected with hereditary hemorrhagic telangiectasia (HHT) presenting with diffuse pulmonary arteriovenous malformations (PAVMs). To correlate the initial presentation and long-term results of embolization according to the distribution of PAVMs. Materials and methods: All consecutively treated patients were divided into three groups, according to the involvement ofevery subsegmental pulmonary artery (group 1), segmental artery (group 2), or both (group 3) of at least one lobe. Age, sex, initial clinical presentation, and Pao2 were recorded before embolization. Per and postprocedural complications were carefully recorded. Clinical outcome and imaging follow-up were obtained at 6 months and annually thereafter. Results: Thirty-nine patients (31 women, 8 men; mean age, 35 years), all of them with bilateral lung involvement, were treated. Group 1 consistedof8, group 2 of17, and group 3 of14 patients. Dyspnea was present in 35 ofthe patients (90%) and cyanosis in 17 patients (44%). Preembolization Pao, was different between groups 1 (52.6 ± 11.6 mm Hg) and 3 (70.7 ± 14.1 mm Hg). Neurologic events were more frequently reported before treatment in group 1 (62.5%) than in group 2 (35%) or in group 3 (43%). Eighty percent of patients reported improvement in their dyspnea after embolization. Pao2 levels improved more in group 2 than in groups 1 and 3. Eight ischemic or infectious complications occurred in 4 patients (10%) due to reperlusion of embolized PAVMs or enlargement of nonembolized PAVMs. Complete and partial treatment success was reported using CT scanning in 59% and 38% of cases, respectively. Conclusion: Dyspnea and paradoxical embolism are frequently encountered in HHT patients with diffuse PAVMs. Prevention of complications and improvement of dyspnea can be achieved after successful embolization in most patients. Better improvement of Pao2 can be achieved in group 2. (CHEST 2009; 135: ) Key words: aneurysm; arteriovenous malformations, embolization; hereditary hemorrhagic telangiectasia, pulmonary artery; transcatheter Abbreviations: HHT = hereditary hemorrhagic telangiectasia; NYHA = New York Heart Association; PAVM = pulmonary arteriovenous malfonnation; TIA = transient ischemic attack pulmonary arteriovenous malformations (PAVMs) are rare direct communications between pulmonary arteries and pulmonary veins, without interposition of a capillary bed. 1,2 These uncommon lesions are usually congenital, with approximately 80 to 90% of patients with PAVMs having hereditary hemorrhagic telangiectasia (HHT).l-3 HHT, also known as Rendu-Osler-Weber syndrome, is an autosomaldominant, multisystem vascular disorder occurring in approximately 10 to 20 individuals per 100, The diagnosis of HHT is definite if at least three of the following criteria (Curacao criteria) are encountered: epistaxis; mucocutaneous telangiectases; family history; and visceral involvement (GI telangiecta- CHEST APRIL,

2 ses, brain, spinal, pulmonary, or liver arteriovenous malformations)." An estimated 15 to 35% of HHT patients present PAVMs.1,2 Some of these patients have a more severe and diffuse pattern of disease. 1.2 The following three types of major complications may be associated with the presence of PAVMs: paradoxical systemic embolism associated with a high risk of stroke, transient ischemic attack (TIA), or abscess; hemoptysis or hemothorax in case ofpulmonary or pleural rupture of PAVMs; and respiratory symptoms such as dyspnea, cyanosis, and clubbing because PAVMs are right-to-ieft shunts. 1,2,.5,6 Management of PAVMs relies mainly on prophylactic administration ofantibiotics before dental and surgical procedures to prevent cerebral abscess and percutaneous transcatheter embolization, which has become the first-line treatment as an alternative to surgery.1,2 Long-term clinical success of embolization in patients with localized PAVMs has been extensively reported.r-" It has been estimated that 7 to 11% of patients present a diffuse form of PAVMs with involvement of several pulmonary segments and lobes. 2,9,10 Diffuse pattern of PAVMs is variably defined as PAVMs involving every segmental or every subsegmental artery of at least one lobe.v!' In a recent report, 10 diffuse involvement was defined as minimally one segment diffusely involved. The number of publications reporting the use of embolization to treat diffuse PAVMs is still very limitedy-ll Our objectives were to review the safety and efficacy of embolization in patients with diffuse PAVMs, to find correlations between initial clinical presentation and the distribution of PAVMs, and to analyze long-term clinical and morphologic results of embolization according to the distribution of PAVMs. MATERIALS AND METHODS Our study was a retrospective analysis of a prospectively acquired database. No specific institutional review board approval is required for this type of retrospective analysis. All "From the Departments of Radiology (Drs. Lacombe. Lagrange, El Hajjam, and Pelage), Statistics and Public Health (Dr. Beauchet), and Pulmonology (Dr. Chinet). H6pital Ambroise Pare, Assistance Publique-Hopitaux de Paris, Universite Paris Ile-dc-France Ouest, Boulogne-Billancourt, France. The authors have reported to the ACCP that no significant conflicts of interest exist with any companies/organizations whose products or services may be discussed in this article. Manuscript receivedjuly2.3, 2CXl8; revision accepted October i, Reproduction of this article is prohibited without written permission from the American College of Chest Physicians ( orgimisc/reprints.shtml). Correspondence to: Jean-Pierre Pelage, MD, PhD, Department of Radiology, Hopital Ambroise Pare, Assistance Puhlique-Hiniitaux de Paris, Univcrsite Paris lle-de-france Ouest, 9 Ave Charles de Gaulle, Boulogne-Billancourt Cedex, France; jeanpierre.pelage@apr.aphpfr DOl: lo.13781chest patients were informed about the risks and benefits of embolization to treat their disease and gave informed consent. All consecutively seen adults and children with an established diagnostic of HHT (with at least three Curacao criteria) and diffuse PAVMs treated with embolization were included. In our study, all patients with diffuse PAVMs were diagnosed with HHT.. Clinical Presentation The following preembolization clinical and biological paraml'- ters were recorded: agl' at presentation, Sl'X; and the presl'ncl' of a genetic mutation. Mutation on two genes, endoglin and ALKl, corresponding to the disease subtypes HHTI and HHT2, respectively, were investigated. Previous pulmonary (hemoptysis, hemothorax) or neurologic complications (stroke, TIA, brain abscess) and previous history of embolization or lung surgery were recorded. The presence of respiratory symptoms, such as dyspnea classified according to the New York Heart Association (NYHA) classification, cyanosis, or clubbing was noted. Pao 2 measurements were obtained before embolization with the breathing room air. Classification of PAVMs For all patients, the lobar, segmental, and subsegmental distribution of PAVMs was recorded according to both CT scan and angiographic findings. The number and architecture of each PAVM (simple or complex) and the diameter of the feeding artery for the dominant PAVM were noted. 12 Three groups of patients were distinguished according to the distribution and number of PAVMs. In group I, PAVMs involved every subsegmental artery of at least one lobe." This means that in the territory of every subsegmental artery there was at least one PAVM. In group 2, PAVMs involved every segmental artery of at least one lobe." In group 3, PAVMs involved every subsegmental artery of at least one lobe and every segmental artery of at least one other lobe. Indications and Technique of Embolization Embolization was indicated for primary or secondary prl'vention of paradoxical embolism or mpture of PAVMs and for improvement of hypoxemia. Conscious sedation with local anesthesia was used in adults, and general anesthesia was favored for children < 14 yl'ars. Prophylactic antibiotics were given 1 h before and for 24 h after embolization, Heparin was also administered. Using a femoral vein approach, a 6F guiding sheath was positioned into till' pulmonary arteries. Pulmonary artery pressures were measured before and after embolization. All patients were embolized using fibered coils (MReye or Nester coils; Cook Medical, Bjaeverskov, Denmark). The technique of embolization consisted of occluding the feeding pulmonary artery as close to the aneurismal sac as possible to prevent reperfusion. As usually recommended, every PAVM with a feeding artery > 3 rum was emholizerl." In areas with a combination of normal vessels and small PAVMs with feeding arteries measuring <.3mm, embolization was performed if the feeding arteries were catheterized successfully. This technique of peripheral redistribution of blood toward the normal pulmonary vessels was named "peripheral blood flow redistribution." The systolic, diastolic, and mean pulmonary artery pressures were recorded before embolization with the transducer leveled to the midaxillary line. Measurements were obtained in spontaneous breathing during 20 consecutive cycles and taken from a di!qtal readout. Pulmonary hypertension W,L~ defined as a mean pressure > 25 mm Hg at rest. Cardiae output was not generally measured Original Research

3 Procedural Complications Technical complications occurring during embolization, including catheterization failure, air or paradoxical embolism, nontarget migration of coils, or vessellpavm rupture, were recorded. Early complications, such as fever, chest pain, pleural effusion, lung infarction or infection, and hemoptysis occurring up to 90 days after embolization, were considered procedure-related. Postemholization Follow-up The first clinical evaluation was c:onducted within 5 months after embolization. The presence of dyspnea classified according to the NYHA classification was noted. Arterial blood gas tests with Pao., evaluation was made and compared with pretreatment levels. Complications occurring > 90 days after embolization were classified as late complications and included paradoxical embolism, hemoptysis, neurologic events, or death. Subsequent clinical evaluation was made every year. The results obtained after the last embolization were used for the final analysis. Imaging follow-up consisted of a contrast-enhanced CT scan of the chest performed at 5 months. The appearance of embolized PAVMs was used to define the success of the procedure, as follows: (1) Success was defined by disappearance or complete retraction of embolized PAVMs without persistent perfusion; (2) partial success was defined by incomplete retraction of the embolized PAVMs with persistent pulmonary and/or systemic perfusion; and (3) failure was defined by unchanged appearance or growth of embolized PAVMs,!3 In the case of incomplete retraction of embolized PAVMs, the type of reperfusion was defined as central recanalization through the coils, recruitment of side pulmonary artery branches, or systemic artery supply (bronchial or nonbronchial systemic arteries).7.11 The presence of new PAVMs or growth of nonembolized PAVMs was also recorded. Subsequent imaging evaluation was obtained at 2 to 5 years in case of successful embolization. In the event offailure or partial success, another embolization was offered to the patient. For analysis, only the comparison between preprocedural imaging findings and those obtained after the last embolization procedure was made. Statistical Analysis seven of them (95%) were adults, and two were children. Twenty-nine patients (74%) presented with endoglin and one (3%) with ALKI mutation. In five patients (15%), no mutation was identified. In the remaining four patients, the results are not yet available. Dyspnea was present in 35 patients (90%) including class II in 15 patients, class III in 12 patients, and class IV in 8 patients. Cyanosis was present in 17 patients (44%), and clubbing in 11 patients (28%). Hemoptysis was reported before embolization in 7 patients (18%). Five patients (13%) underwent previous pulmonary surgical lobectomy, and 5 patients (13%) underwent a previous failed embolization of PAVMs at other institutions. Preembolization Pa02 measurements were 52.6 ± 11.6,65.5 ± 18.2, and ± 13.8 mm Hg, respectively, in groups 1, 2, and 3. There was a significant difference between groups 1 and 3 (p < 0.05). The initial Pao, was also related to the diameter of the artery of the dominant PAVM (p = ) and to the number of segments with innumerable PAVMs (p = 0.037). There were no correlation between the initial Pao, and the number of involved or noninvolved lobes. Neurologic complications occurred before embolization in 17 of 39 patients (44%). Stroke, TIA, and brain abscess were reported in 4, 4, and 9 patients, respectively. Neurologic events were more frequently reported in group 1 (62.5%) than in group 2 (35%) or in group 3 (43%) (p < 0.05). Classification ofpavms According to the distribution of PAVMs, there were 8 patients (20%) in group 1, 17 patients (44%) Quantitative parameters were expressed as mean ± SD and range. Qualitative data were given in frequency and percentage. Analysis of variance was obtained in case of significance by multiple comparisons tests using the contrast method. Multiple regression analysis was used to evaluate the relationship between significant criterion and predictor variables. A p value < 0.05 was considered statistically significant. A univariate and then a multivariate analysis was conducted to look for predictive factors of treatment failure (reperfused PAVMs on follow-up CT scan after embolization). Statisticalanalyseswere performed using a statistical software package (SAS, version 8.2; SAS Institute; Cary, NC). RESULTS Demographics and Initial Presentation From February 1992 to June 2007, a total of 39 patients (31 women and 8 men) with a mean age of 35 ± 16 years (age range, 9 to 83 years; median age, 30 years) with diffuse PAVMs and confirmed diagnosis of HHT were treated with embolization. Thirtywww.chestjoumal.org FIGURE 1. Imaging of a 55-year-oldwoman with HHT belonging to group 1. Left, A: CT scan (multiplanar coronal reconstruction) with lung parenchyma window settings showing multiple PAVMs involving every subsegmental artery of the left lower lobe. Right, B: selective angiography of the left lower lobe pulmonary artery (performed at the time of embolization) confirms the classification (subsegmental involvement). CHEST/135/4/ APRIL,

4 FIGURE 2. A CT scan (multiplanar sagittal reconstruction) of a 72-year-old woman with HHT belonging to group 2 showing lung parenchyma window settings and demonstrating multiple PAVMs involving every segmental artery of the left lower lobe. in group 2, and 14 patients (36%) in group 3 (Fig 1, 2). PAVMs were bilateral in all patients. The right lung (59%) was more frequently involved than the left lung (41%). Both lower lobes were involved in all patients; the middle lobe (95%), the right upper lobe (92%), and the left upper lobe and lingula (74%) were less frequently affected. A simple type or a complex type of PAVMs was encountered in 34 of 39 patients (87%) and 5 of 39 patients (13%), respectively. A total of 443 PAVMs had a diameter 2: 3 mm. The artery of the dominant PAVM measured 6.1 ± 2.1 mm (range, 3 to 12 mm). Embolization Procedures Seventy-one procedures were performed (mean, 1.8; range, 1 to 9). The number of embolization sessions was not significantly different between the three groups of patients. A total of 681 PAVMs (mean, 17) were embolized. This included 443 PAVMs with a feeding artery 2: 3 mm and 238 PAVMs with a feeding artery :::;; 3 mm treated using the peripheral blood flow redistribution technique. The number of embolized PAVMs was significantly different between group 1 (36.9 ± 41.9) and group 2 (6.4 ± 4.1; P < 0.05). A total of 1,151 coils (mean, 29 coils per patient) were used to occlude the feeding artery of PAVMs. A pulmonary hypertension was initially present in 2 of 33 patients (6%). In these two patients, the mean pulmonary artery pressure remained unchanged between the first and the last session. After the first embolization, the systolic pulmonary pressures increased from a mean of 21 to 23.3 mm Hg; the mean pulmonary artery pressure increased from a mean of 13.6 to 15.3 mm Hg. The average values of systolic and mean pulmonary artery pressures remained unchanged in repeated embolizations. Complications of Embolization A total of 21 procedural or early complications were reported in 20 patients. Table 1 provides the details. Seven complications occurred during embolization in 6 patients, and 14 occurred during the first 2 months postembolization in 14 patients. No paradoxical migration of coil through the PAVM occurred during embolization. Of note, catheter-induced arterial wall damage occurred during catheterization of the feeding artery in two patients, both of them with pulmonaryhypertension. One of these two patients presented a mild hemoptysis successfully treated with prompt embolization. Clinical Results Initial clinical follow-up (within the first 6 months postembolization) was available in all patients, but one patient was lost to follow-up 12 months after embolization. The mean follow-up duration was 43.3 months (range, 4 to 184 months). Three patients died from cerebral hemorrhage, myocardial infarction, and a cerebral abscess, respectively. Respiratory symptoms were precisely evaluated in 35 patients. Improvement was noted in 28 of 35 Table I-List ofcomplications Onset of Complications Type of Complications Description Coil reflux Failed catheterization Arterial wall damage TIA Sudden deafness Fever Pulmonary infarction Pleural affusion Lung abscess No. Treatment Required Early Delayed Total Technical Clinical Clinical 2 I I 21 None Retreatment Immediate embolization None None Symptomatic Symptomatic Drainage Antibiotics 1034 Original Research

5 Table 2-Pao 2 Measurements Before and After Embolization Pretreatment Measurement, Posttreatment Measurement, Gain After Embolization, mm Hg mm Hg mmhg Groups Patients, No. IMean :!: SD Median I IMean :!: SD Median I I Mean :t: SD Median I :t: :!: :t: :t: :!: :t: :!: :!: :t: patients (80%). A gain of one NYHA grade was observed in 19 of35 patients (54%), of two grades in 7 of35 patients (20%), and of three grades in 2 of35 patients (6%). There was no improvement in 7 of 35 patients (20%). Table 2 presents the Pao., measurements obtained before the first and after the last embolization. A significant difference was noted between the three groups (p < ), between groups 1 and 2 (p = 0.002), and between groups 1 and 3 (p = 0.002). Pao2 levels improved more in group 2 than in groups 1 and 3. A total of eight ischemic or infectious complications occurred in four patients (10%) during the follow-up. All of them had either a reperfusion of an embolized PAVM or an enlargement of nonernbolized PAVMs. One woman died from a cerebral abscess and acute respiratory failure. Another woman experienced a TIA and a brain abscess 1 week after a cesarean section performed without prophylactic antibiotics. Two episodes of TIA occurred in one patient with enlargement of nonembolized PAVMs. A septic hip arthritis and a TIA occurred in another patient with reperfusion of a complex PAVM. Seven full-term pregnancies (four vaginal deliveries and three cesarean sections) occurred in six women. Worsening of respiratory symptoms with enlargement of nonembolized PAVMs was reported after five pregnancies in four women, whereas clinical condition remai.ned stable in two women. Imaging Follow-up Imaging evaluation after the last embolization was available in 34 patients. The remaining five patients had been embolized recently, and CT scanning had not yet been performed. A successful treatment of all embolized PAVMs was observed in 20 patients (59%). A partially successful treatment of embolized PAVMs was noted in 13 patients (38%), and 1 patient (3%) did not respond to treatment due to catheterization failure. The partially successful failures (nonretracted PAVMs) consisted of nine cases of central recanalization, two cases of collateral side-branch recruitment, and two cases of systemic artery supply. Univariate analysis failed to demonstrate any predictive factor for failure so that no multivariate analysis was conducted. A stability of nonembolized PAVMs was noted in 24 patients (70%), and enlargement was seen in 10 patients (30%). In four cases, enlargement was detected after a pregnancy. No new PAVM was seen on a CT scan. DISCUSSION The presence of diffuse PAVMs is reported in < 10% of patients with hereditary PAVMs.2 Most patients with diffuse PAVMs present a more severe hypoxemia and a higher rate of neurologic complications than those with localized PAVMs.9.14,15 Surprisingly, there is currently no consensus regarding the definition of diffuse PAVMs. Some authors have defined them as the involvement of every segmental artery of at least one lobe, and others have defined them as the involvement of every subsegmental artery of at least one lobe, or more recently a diffuse involvement of at least one segment.1,2,9-11 Therefore, based on the definitions proposed in the literature, we have distinguished three groups of patients. Group 1 consisted of patients with involvement of every subsegmental artery of at least one lobe, and group 2 consisted of patients with involvement of every segmental artery of at least one lobe. Patients belonging to group 3 had a combination of subsegmental and segmental involvement of at least one lobe. We found that neurologic complications at presentation were more frequent in group 1 (62.5%) than in the two other groups. This percentage is close to the those (70%) already reported by Faughnan et al 9 and by Pierucci et al.!? This may indicate that the severity ofthe disease was similar between the patients ofour group 1 and the adults described by Faughnan et al? and the group of patients with bilateral involvement reported in study by Pierucci et al.'? We have also found that Pao, measurements were lower in group 1 than in the two other groups. Of note, patients belonging to group 3 had the highest Pao, level before embolization, whereas they had by definition at least two lobes involved. This may be due to differences in terms of affected lobes. For instance, CHEST/135/4/ APRIL,

6 some patients may have only the middle lobe affected, and others, two lobes of smaller volume. Surprisingly the severity of respiratory symptoms was lower and the levels of Pao, higher before treatment in the two groups described by Pierucci et al,1o but some patients were children and had unilateral PAVMs, whereas all our patients were adults with bilateral involvement. In our study, the initial level of Pao., was correlated with the diameter of the dominant PAVM.14 It is usually considered that the main objective of embolization is the primary or secondary prevention of paradoxical embolism. Four patients of our study presented late ischemic and/or septic complications. The ischemic events (TIAs) occurred in patients with either reperfusion of embolized PAVMs or enlargement of nonembolized PAVMs. Our rate of late complications is similar to that reported by other groups. 1,9 In a series of 42 children with focal (71%) or diffuse (29%) PAVMs treated with embolization, 45% of patients presented reperfusion of PAVMs and most of them had to be retreated.1 1,16-20 The reperfusion rate is similar to ours (38%), even if it is considered that children may develop collateral vessels more commonly than adults.'! In patients with diffuse PAVMs, it may be advised to conduct annual follow-up for early detection of reperfusion or growing PAVMs, Embolization is currently recommended for every PAVM with a feeding artery :::: 3 mm. However, embolization ofthe large PAVMs only is known to be insufficient to improve hypoxemia because many other small PAVMs remain patent." The pulmonary flow redistribution consists of performing a permanent proximal pulmonary embolization in diffusely involved areas." This technique has been applied in addition to the more conventional embolization technique in 12 of 14 patients with diffuse PAVMs.9 Unfortunately, most of these patients did not report any clinical improvement. The peripheral blood flow redistribution we have chosen allows segmental blood flow redistribution toward normal pulmonary arteries. After treatment using conventional embolization alone or in combination with peripheral blood flow redistribution, a significant improvement of respiratory symptoms and Pao, was reported in 80% of patients. These results seem better than those previously reported, but it should be emphasized that the increase in Pao., was Significantly lower in patients of group 1 (subsegmental involvement) than in patients of the two other groups.9 There are some limitations to this study. Since the firstembolization performed in 1992, our imagingmodalities and embolization techniques have changed.l 8,19 We report a mean follow-up duration of 4 years, which may be too short to draw definitive conclu sions on the rate of reperfusion of embolized PAVMs.l6,18 Again, annual follow-up may be advised even if radiation exposure should be taken into consideration, particularly for young affected patients. In conclusion, it is well known that hypoxemia, hemoptysis, and neurologic symptoms are frequently encountered in HHT patients with diffuse PAVMs. Our experience suggests that patients should be divided into three groups according to the subsegmental or segmental distribution of PAVMs. Paradoxical embolism is more frequently encountered in patients with involvement of every subsegmental artery of at least one lobe. Prevention ofparadoxical embolism and improvement of dyspnea can be achieved after successful embolization in most patients. However, referring physicians should be aware that better improvement of Pao, can be expected in patients with involvement of every segmental artery of at least one lobe compared to those with a more peripheral involvement (subsegmental). This classification of patients with diffuse PAVMs seems to be clinically relevant to define long-term prognosis. ACKNOWLEDGMENTS: We would like to thank the following participating physicians of the Paris HHT center, Hopital Ambroise Pare, Universite Paris Ouest: I. Bourgault, MD, Department of Dermatology; J. H. Blondel, MD, Ear, Nose, and Throat Department; J. C. Farcot, MD, Department of Cardiology; G. Lesur, MD, Department of Hepatogastroenterology; B. Raffestin, MD, Functional Exploration Laboratory; and J. Roume, MD, Department of Genetics, Hopital Ambroise Pare, Assistance Publique Hopitaux de Paris, Universite Paris Ile-de-France Ouest, Boulogne, France. We also thank A. Ozanne, MD, Department of Neuroradiology, and P. Lasjaunias (in memoriam), Department of Neuroradiology, Centre Hospitalier de Bicetre, Assistance Publique Hopitaux de Paris, Universite Paris Sud, Le Krernlin-Bicctre, France. REFERENCES 1 Burke CM, Safai C, Nelson DP, et al. Pulmonary arteriovenous malformations: a critical update. Am Rev Respir Dis 1986; 134: Gossage JR, Kan] G. Pulmonaryarteriovenous malformations: a state of the art review. Am J Respir Crit Care Med 1998; 158: Guttmacher AE, Marchuk DA, White RI. Hereditary hemorrhagic telangiectasia. N Engl J Med 1995; 333: Shovlin CL, Guttmacher AE, Buscarini E, et al. Diagnostic criteria for hereditary hemorrhagic telangiectasia (Rendu- Osier-Weber syndrome). Am J Med Genet 2000; 91: Hewes RC, Auster M, White RI. Cerebral embolism: first manifestation of pulmonary arteriovenous malformations in patients with hereditary hemorrhagic telangiectasia. Cardiovasc Intervent Radiol 1985; 8: Ference BA, Shannon TM, White RI, et al. Life-threatening pulmonary hemorrhage with pulmonary arteriovenous malformations and hereditary hemorrhagic telangiectasia. Chest 1994; 106: White RI, Lynch-Nyhan A, Terry P, et al. Pulmonary arte- Original Research

7 riovenous malformations: techniques and long-term outcome of ernbolotherapy. Radiology 1988; 169: White Rl, Pollak JS, Wirth JA. Pulmonary arteriovenous malformations: diagnosis and transcatheter ernbolotherapy. J Vasc Interv Radiol 1996; 7: Faughnan ME, Lui YW, Wirth JA, et al. Diffuse pulmonary arteriovenous malfonnations: characteristics and prognosis. Chest 2000; 117: Pierucci P, Murphy J, Henderson KJ, et al. New definition and natural history of patients with diffuse pulmonary arteriovenous malformations: twenty-seven-year experience. Chest 2008; 133: Faughnan ME, Thabet A, Mei-Zahav M, et al. Pulmonary arteriovenous malformations in children: outcomes of transcatheter embolotherapy. J Pediatr 2004; 145: White Rl, Mitchell SE, Barth KH, et al. Angioarchitecture of pulmonary arteriovenous malformations: an important consideration before embolotherapy. AJR Am J Roentgenol1983; 140: Remy-Jardin M, Remy J, Artaud D, et al. Peripheral pulmonary arteries: optimization of the spiral CT acquisition protocol. Radiology 1997; 204: Lee DW, White Rl, Egglin TK, et al. Embolotherapy oflarge pulmonary arteriovenous malformations: long-term results. Ann Thorac Surg 1997; 64: Dutton JA, Jackson JE, Hughes JM, et al. Pulmonary arteriovenous malformations: results of treatment with coil embolization in 53 patients. AJR Am J Roentgenoll995; 165: Mager JJ, Overtoom TT, Blauw H, et al. Embolotherapyof pulmonary arteriovenous malformations: long-term results in 112 patients. J Vase Interv Radiol 2004; 15: Millie A, Chan RP, Cohen JH, et al. Reperfusion of pulmonary arteriovenous malformations after embolotherapy. J Vase Interv Radiol 2005; 16: Remy-Jardin M, Dumont P, Brillet PY, et al. Pulmonary arteriovenous malformations treated with embolotherapy: helical CT evaluation of long-term effectiveness after year follow-up. Radiology 2006; 239: Brillet PY, Dumont P, Bouaziz N, et al. Pulmonary arteriovenous malformation treated with embolotherapy: systemic collateral supply at multidetector CT angiography after year follow-up. Radiology 2007; 242: Lacombe P, Lagrange C, EI Hajjam M, et al. Reperfusion of complex pulmonary arteriovenous malformations after embolization: report of three cases. Cardiovasc Intervent Radiol 2005; 28: CHEST/135/4/ APRIL,