Intrapulmonary right-to-left shunting detectable by saline

Incorporation of the Hepatic Veins Into the Cavopulmonary Circulation in Patients With Heterotaxy and Pulmonary Arteriovenous Malformations After a Kawashima Procedure Doff B. McElhinney, MD, Jacqueline Kreutzer, MD, Peter Lang, MD, John E. Mayer, Jr, MD, Pedro J. del Nido, MD, and James E. Lock, MD Departments of Cardiology and Cardiac Surgery, Children s Hospital, and Departments of Pediatrics and Surgery, Harvard Medical School, Boston, Massachusetts Background. In patients with polysplenia syndrome and azygous continuation of an interrupted inferior vena cava (IVC), pulmonary arteriovenous malformations (PAVMs) are relatively common after bidirectional cavopulmonary anastomosis (BCPA, Kawashima procedure). Resolution of PAVMs after hepatic vein (HV) inclusion into the cavopulmonary circulation has been reported, but there has been no systematic investigation of the effects of this therapy in a population of more than 3 patients. Methods. We studied 16 patients with heterotaxy, univentricular congenital heart disease, and azygous continuation of the IVC who underwent incorporation of the HV into the cavopulmonary circuit for treatment of significant PAVMs after a Kawashima procedure. Results. The median preoperative systemic arterial oxygen saturation (S sa O 2 ) was 76% (65% 85%), compared with 89% (85% to 92%) early after BCPA. Among 15 early survivors, the median early postoperative S sa O 2 was 76% (56% 85%). In 11 of the 15 survivors, S sa O 2 rose to 90% or greater within a year and remained at 93% or greater at follow-up of 2.8 to 10 years. Four patients had persistent hypoxemia and residual PAVMs at follow-up catheterization 1.5 to 8 years postoperatively; these patients had the most severe hypoxemia prior to HV inclusion, and in 2 the residual PAVMs were unilateral, with HV flow streaming to the contralateral lung, in which PAVMs had resolved. Conclusions. Hypoxemia resolved after cavopulmonary incorporation of the HV in the majority of our patients with PAVMs after the Kawashima operation, presumably due to a combination of PAVM resolution and elimination of hepatic venoatrial right-to-left shunting. These findings support the theory that development of PAVMs is facilitated by exclusion of HV effluent from the pulmonary circulation. (Ann Thorac Surg 2005;80:1597 603) 2005 by The Society of Thoracic Surgeons Intrapulmonary right-to-left shunting detectable by saline contrast echocardiography or radionuclide lung perfusion scintigraphy is present almost universally after superior cavopulmonary anastomosis (CPA) [1 5]. However, pulmonary arteriovenous malformations (PAVMs) causing significant hypoxemia after a classic Glenn procedure (ie, right CPA) or bidirectional CPA (BCPA) occur in only a small minority of patients, with increasing prevalence according to the duration of time since the CPA was created [1, 6 8]. In the current era of system- See page 1604 atically staged univentricular palliation, the classic Glenn procedure has been replaced by BCPA, which is typically converted to a Fontan connection within 1 to 3 years. Accordingly, the duration of time a patient remains with Accepted for publication May 9, 2005. Address correspondence to Dr McElhinney, Department of Cardiology, Children s Hospital, 300 Longwood Avenue, Boston, MA 02115; e-mail: doff.mcelhinney@cardio.chboston.org. a CPA tends to be limited and clinically important PAVMs are uncommon. An exception to this pattern is the population of patients with the polysplenia form of heterotaxy in whom the inferior vena cava (IVC) is interrupted and lower body venous blood drains to the superior vena cava (SVC) through an azygous or hemiazygous vein. Such patients develop PAVMs after a Kawashima procedure (BCPA in the setting of azygous continuation of an interrupted IVC to the SVC) [9] frequently and acutely in some cases [10 16]. The cause of PAVMs is unknown. However, there is increasing evidence to support the hypothesis that development of PAVMs is facilitated when an unidentified factor produced in the liver does not reach the pulmonary circulation [10]. In individuals with congenital heart disease, PAVMs develop almost exclusively in circumstances in which hepatic vein (HV) blood does not perfuse the pulmonary circulation before first traversing a systemic capillary bed [1 5, 8, 10]. PAVMs are also a feature of the hepatopulmonary syndrome that can occur in patients with hepatic cirrhosis [17], and of hereditary 2005 by The Society of Thoracic Surgeons 0003-4975/05/$30.00 Published by Elsevier Inc doi:10.1016/j.athoracsur.2005.05.101

1598 MCELHINNEY ET AL Ann Thorac Surg PULMONARY ARTERIOVENOUS MALFORMATIONS 2005;80:1597 603 hemorrhagic telangiectasia (HHTA) [18]. The hepatic factor hypothesis is supported by reports of PAVMs resolving in patients with a BCPA after diversion of HV effluent to the pulmonary arteries (PAs) [11, 14, 19 21] or cardiac transplant [22], or after liver transplant in patients with the hepatopulmonary syndrome [23]. Whether such reports are unique cases or the typical response to direction of HV effluent to the pulmonary circulation, however, is unclear, as there has been no systematic investigation of the clinical course after cavopulmonary incorporation of the HV in patients with univentricular heart disease who have developed PAVMs after BCPA. Patients and Methods Our intent was to assess the effect of directing HV effluent to the pulmonary circulation in patients with univentricular congenital heart disease who had developed clinically important PAVMs after CPA. In order to standardize the patient population and minimize the potentially confounding effects of other pulmonary vascular changes that can occur in the context of a unidirectional CPA or acquired discontinuous PAs, we limited the study to patients with heterotaxy, univentricular heart disease, azygous or hemiazygous (hereafter, azygous ) continuation of an interrupted IVC, confluent PAs, and direct HV drainage to the atrium, and who developed PAVMs after a Kawashima procedure. Diagnostic Criteria Pulmonary arteriovenous malformations were diagnosed when there were the following: (1) rapid pulmonary AV transit ( 3 heart beats) of contrast on proximal PA angiography; (2) a typical reticular or spongy pattern in the peripheral pulmonary vasculature on PA angiography; and (3) for the 13 patients in whom pulmonary venous (PV) blood was sampled, PV desaturation ( 92%) in angiographically affected lung segments. Data Analysis Within-patient and between-patient data were compared using paired and independent samples t test analyses, respectively. Repeated measures analysis of variance was used to compare serial measurements between groups of patients. Unless otherwise specified, oxygen saturations were measured in room air. Data are expressed as mean standard deviation or median (range). Control Patients Clinical management and outcomes were reviewed for a control group of patients with heterotaxy, univentricular heart disease, and azygous continuation of an interrupted IVC to the SVC who underwent BCPA and/or total cavopulmonary connection at our center between 1985 and 2001 and did not have PAVMs. Results Patients Between 1990 and 2001, 16 patients with functionally univentricular heart disease, azygous continuation of an interrupted IVC, and a Kawashima procedure were diagnosed with clinically important PAVMs at cardiac catheterization and underwent surgical intervention to direct HV flow to the pulmonary circulation. Systemic venous anatomy was notable for bilateral SVCs in 10 patients, a single left SVC in 4, and a single right SVC in 2, with the interrupted IVC returning through the ayzgous system to the right SVC in 8 patients and to the left SVC in 8. Surgical procedures prior to BCPA had been performed in 10 patients, primarily to augment (n 9) or limit (n 2) pulmonary blood flow, or to augment the ascending aorta and aortic arch (n 3). Details of 5 of these patients prior to HV incorporation were included in a previous report [10]. Bidirectional Superior Cavopulmonary Anastomosis A Kawashima procedure had been performed in all patients (14 at our center and 2 elsewhere) between 1985 and 1999, at a median age of 10 months (4 months to12 years). Nine of 10 patients with bilateral SVCs underwent bilateral BCPA; in 1 patient, a nondominant left SVC was ligated. The median systemic arterial oxygen saturation (S sa O 2 ) early after BCPA was 89% (85% to 92%; Fig 1). Diagnosis of PAVMs The diagnosis of PAVMs was made at catheterization a median of 3.1 years (6 months to 10 years) after BCPA. The PAVMs were diagnosed by angiography in all patients, with bilateral PAVMs in 15 and unilateral PAVMs in 1. Two patients had large, angiographically discrete PAVMs in addition to the typical diffuse PAVMs seen in all patients (Fig 2). In all 13 patients in whom PV blood sampling was performed (median 2 PVs sampled), there was PV desaturation ( 92%) in at least 1 PV (median 2 PVs desaturated) corresponding with angiographic PAVMs, without correction in 100% oxygen. In the patient with unilateral angiographic PAVMs, there was PV desaturation ipsilateral to the angiographic PAVMs and normally saturated PV blood from the lung without angiographic PAVMs. The S sa O 2 at catheterization ranged from 65% to 85% (median 76%; p 0.001 vs early post-bcpa measurement; Fig 1), and the median hemoglobin level was 18.0 g/dl (15.2 19.7 g/dl). The patient with the highest S sa O 2 (85%) was the only patient with unilateral PAVMs. Eight patients underwent coil embolization of systemic-pulmonary venous collaterals during the same catheterization. Cavopulmonary Incorporation of the HV The HV were directed to the cavopulmonary circuit with an intraatrial lateral tunnel in 7 patients, an extracardiac conduit (13 19 mm) from the HV to the cavopulmonary circuit in 8, and an intracardiac conduit in one. In 8 patients, the baffle or conduit was fenestrated.

Ann Thorac Surg MCELHINNEY ET AL 2005;80:1597 603 PULMONARY ARTERIOVENOUS MALFORMATIONS 1599 patients had persistent hypoxemia, with a S sa O 2 from 69% to 79% at most recent follow-up (4.4 to 10 years). Among the entire cohort, S sa O 2 was 90 9% at follow-up vs 76 8% in the early postoperative period and 74 7% at preoperative catheterization (p 0.001). Age, duration between BCPA and diagnosis of PAVMs, and S sa O 2 after BCPA were similar in these patients and those in whom hypoxemia resolved. However, the 4 patients with persistent hypoxemia had significantly lower S sa O 2 both at the time PAVMs were diagnosed and at the time of hospital discharge after HV incorporation. They also had a significantly greater decrease in S sa O 2 between BCPA and the diagnosis of PAVMs (p 0.001; Fig 1). In all 4 of these patients, including both of those with large, angiographically discrete PAVMs preoperatively, and one with a patent conduit fenestration, persistent hypoxemia was documented by cardiac catheterization to be due at least in part to PAVMs. Fig 1. Line graph depicting systemic arterial oxygen (SaO 2 ) saturation early after BCPA, at the time PAVMs were diagnosed, at the time of hospital discharge after HV inclusion, and at most recent follow-up. Solid symbols ( ) and lines represent patients with improvement of hypoxemia at most recent follow-up, and open symbols ( ) with dashed lines represent patients with persistent hypoxemia and PAVMs. (BCPA bidirectional superior cavopulmonary anastomosis; HV hepatic vein; PAVMs pulmonary arteriovenous malformations.) Follow-Up Catheterization Follow-up catheterization was performed in 7 patients (13 catheterization procedures), including 3 of the 11 with stable S sa O 2 93% or greater and all 4 with persistent hypoxemia. In the 3 catheterized patients with stable S sa O 2 93% or greater, there was angiographic resolution of PAVMs at the most recent catheterization and normal PV saturations in the previously affected lung(s). One of these patients had undergone prior catheterization 1 year postoperatively, when the baffle fenestration re- Early Outcomes One patient died 26 days after HV incorporation, with severe cyanosis, acute liver failure, low cardiac output secondary to poor ventricular function, and ultimately multiorgan failure. This patient had undergone atrioventricular valve replacement at the time of HV incorporation and pacemaker placement in the early postoperative period for heart block. In 9 of the 15 survivors, the early postoperative course was notable for severe hypoxemia (S sa O 2 65%, usually on supplemental oxygen) lasting at least a week. Five of these 9 patients had a fenestration, while 2 of the 6 patients without severe hypoxemia had a fenestration (p not significant). Two patients developed hemidiaphragm paralysis, one bilaterally and one on the side of the HV pathway. The median duration of hospitalization after HV incorporation was 10 days (4 to 34 days). The median S sa O 2 at the time of hospital discharge was 77% (56% to 90%; p 0.26 vs preoperative). Two patients were discharged on supplemental oxygen. Follow-Up All 15 early survivors were alive at a median crosssectional follow-up of 6.6 years (2.8 to 10 years). Systemic arterial oxygen saturation rose to greater than 90% between 1 week and 12 months (median 4 months) postoperatively in 11 of the 15 early survivors. At a median follow-up of 6.3 years (2.8 to 10 years), these patients, including one with an open fenestration, maintained stable S sa O 2 93% or greater (median 95%). The other 4 Fig 2. Left pulmonary artery angiogram 2 years after incorporation of the hepatic vein (HV) into the Kawashima circulation in a patient with large, angiographically discrete pulmonary arteriovenous malformations bilaterally that did not resolve after HV inclusion.

1600 MCELHINNEY ET AL Ann Thorac Surg PULMONARY ARTERIOVENOUS MALFORMATIONS 2005;80:1597 603 Fig 3. Angiograms (A) 11 months after HV incorporation, prior to fenestration closure, and (B) 4 years after HV incorporation in a patient with bilateral PAVMs in whom there was substantial transfenestration flow of HV blood and persistent PAVMs 11 months postoperatively (A), and resolution of PAVMs after fenestration closure (B). In both images, the catheter course is antegrade through the right SVC, across the central pulmonary artery, and retrograde up the left SVC, with contrast injection in the left SVC. Although these images only demonstrate PAVMs (A) and their resolution (B) in the left lung, PAVMs were present and subsequently resolved bilaterally. (HV hepatic vein; PAVMs pulmonary arteriovenous malformations; SVC superior vena cava.) mained patent and the majority of HV flow was through the fenestration, at which time the PAVMs were still present (Fig 3). The fenestration was closed, and at subsequent catheterization 3 years later the PAVMs had resolved (Fig 3). In all 4 patients with persistent hypoxemia, PAVMs were still present at follow-up catheterization 1.5 to 8 years postoperatively. Two of these patients had large, angiographically discrete PAVMs preoperatively, which were still present at follow-up catheterization (Fig 2). In the other 2, who had bilateral PAVMs prior to HV inclusion, the PAVMs identified at late catheterization were unilateral, with all or most HV flow streaming to the contralateral lung (Fig 3), in which the PAVMs had resolved. Reinterventions Interventions for PAVMs were performed in all 4 patients with significant residual hypoxemia, 1.7 to 5.9 years after HV incorporation, including (1) coil embolization of PAVMs in 3 patients (7 procedures), and (2) creation of a brachial AV fistula ipsilateral to the PAVMs in 2 patients with unilateral PAVMs and streaming of HV flow to the contralateral lung (Fig 4). The S sa O 2 increased 3% to 13% acutely after embolization of PAVMs, but the improvements were transient and all 3 patients remained hypoxemic. There was no lasting S sa O 2 improvement in the patients treated with a brachial AV fistula. Two patients had stenting of the central PAs between bilateral SVCs. In one of these patients, there was streaming of HV flow to the left lung prior to stenting, with resolution of left-sided PAVMs and persistence of right lung PAVMs; follow-up angiography after PA stenting revealed improved flow of HV blood to the right lung, but did not result in resolution of PAVMs. Two patients underwent device closure of the Fontan fenestration at the time of follow-up catheterization. Control Patients In addition to the 16 study patients, 39 patients with heterotaxy, univentricular heart disease, and azygous Fig 4. Angiogram in the HV-PA pathway 7 years after incorporation of the HV into a Kawashima circulation, demonstrating streaming of HV flow to the left lung, in which PAVMs had resolved. The PAVMs persisted in the right lung (not shown). The catheter was passed antegrade into the right SVC through the azygous vein, across the cavopulmonary anastomosis, and retrograde into the HV-PA pathway. (HV hepatic vein; PA pulmonary artery; PAVMs pulmonary arteriovenous malformations; SVC superior vena cava.)

Ann Thorac Surg MCELHINNEY ET AL 2005;80:1597 603 PULMONARY ARTERIOVENOUS MALFORMATIONS continuation of an interrupted IVC to the SVC underwent BCPA and/or total cavopulmonary connection at our center between 1985 and 2001. A BCPA was performed in 20 of these patients at a median age of 1.7 years (3 months to 20 years; p 0.40 vs study patients), and 19 underwent total cavopulmonary connection including the HV, without intermediate BCPA, at ages 2.1 to 15 years (median 4.0 years). Of the 20 patients who underwent BCPA, one died early after BCPA, 2 died less than 2 years post-bcpa without undergoing HV inclusion, 5 were followed elsewhere, 2 were awaiting HV inclusion at our center, and 10 underwent HV inclusion 1.0 to 12 years after BCPA (median 3.0 years, p 0.64 vs study patients). At catheterization prior to HV inclusion in these 10 patients, the median S sa O 2 was 88% (77% to 92%, p 0.001 vs study patients), and none had PAVMs. Comment Effect of Incorporating HV Effluent Into the Cavopulmonary Circulation on PAVMs The purpose of this study was to determine the effect of directing HV effluent to the PAs in 16 patients with heterotaxy who developed progressive hypoxemia due to PAVMs after a Kawashima procedure. In the majority of our patients, hypoxemia resolved after cavopulmonary incorporation of the HV, with stabilization of S sa O 2 at 93% or greater at follow-up of 2.8 to 10 years. Although clinical improvement was noted in 11 of 15 surviving patients, only 3 of these 11 patients underwent follow-up catheterization to confirm resolution of PAVMs. Thus, while resolution of PAVMs may be inferred from substantially improved S sa O 2 in the other 8 patients, corroborating catheterization data were not obtained. Nonetheless, the clinical issue is not so much whether PAVMs resolve anatomically after HV inclusion as whether the physiologic impairment (ie, hypoxemia) is corrected. The clinical improvement that occurred in the majority of our patients was likely due to a combination of factors, including elimination of both intrapulmonary right-toleft shunting through PAVMs and hepatic venoatrial right-to-left shunting. There are a number of reasons to believe that increased S sa O 2 after HV inclusion was due at least in part to decreased shunting through PAVMs. First, PAVMs and their contribution to preoperative hypoxemia were documented in all patients, and S sa O 2 prior to HV incorporation was significantly lower in study patients than a comparable cohort of control patients with a Kawashima procedure who did not have PAVMs. Although S sa O 2 typically decreases with age after a BCPA, most likely due to redistribution of systemic blood flow to the lower body [24], the magnitude of this decrement is not commensurate with the severity of hypoxemia observed in our patients. Second, in 3 patients, resolution of PAVMs was documented at catheterization. Third, the majority of patients had S sa O 2 93% or greater at follow-up, which is unlikely in the context of significant persistent PAVMs. 1601 Fourth, the time course of improved S sa O 2 after HV inclusion was inconsistent with the explanation that hypoxemia improved simply because hepatic venoatrial right-to-left shunting was eliminated: early postoperative S sa O 2 was very low in most patients, improving by the time of hospital discharge to preoperative levels, and only gradually rising above 90% over the next month to year. Elimination of hepatic venoatrial right-to-left shunting alone should have improved hypoxemia immediately after redirection of HV flow, whereas resolution of PAVMs might be expected to occur gradually. Although hypoxemia resolved after HV inclusion in the majority of our cohort, S sa O 2 never normalized in 4 patients with persistent PAVMs at follow-up catheterization. These patients differed from those in whom hypoxemia resolved in several respects: (1) they had the most severe hypoxemia at the time of PAVM diagnosis; (2) 2 of them had large, angiographically discrete PAVMs before HV inclusion; and (3) 2 others had streaming of flow in the HV-PA pathway such that HV blood flowed exclusively or primarily to one lung, in which PAVMs had resolved, while PAVMs persisted in the lung not receiving HV flow. Unilateral streaming of HV-PA flow is an important concern after HV inclusion in patients with heterotaxy and azygous continuation of the IVC. Even if bilateral SVCs are present, the SVC receiving azygous return carries at least 50% of upper body venous return and essentially all lower body and nonsplanchnic abdominal venous return. If this SVC is contralateral to or widely offset from the HV-PA pathway, there is no mechanical impetus for HV effluent to flow to the lung on that side. As investigators have shown using in vitro and computational simulations, flow streaming may be mechanically advantageous in a total cavopulmonary circulation, and offset between the IVC-PA and SVC-PA connections can produce complete or near-complete streaming of HV effluent to the ipsilateral lung [25]. Etiology of PAVMs After a Kawashima Procedure Patients with the polysplenia form of heterotaxy seem to be more prone to develop PAVMs and to develop them more rapidly after a Kawashima procedure, in which all systemic venous return except HV and coronary venous effluent is to the PA circulation, than patients with other forms of heart disease who undergo BCPA [4, 10, 11, 13 16]. Patients with heterotaxy and polysplenia have been reported to develop PAVMs without a Kawashima circulation, but only in the setting of biliary atresia, which is associated with the hepatopulmonary syndrome in its own right [10, 17, 23]. It may be the combination of heterotaxy and a Kawashima circulation that predisposes patients to the formation of PAVMs. Insights into the pathogenesis of PAVMs in patients with a CPA, particularly those with heterotaxy and a Kawashima circulation, may be offered by the genotypic characterization of HHTA, in which patients develop PAVMs and other vascular anomalies. The majority of patients with HHTA are haploinsufficient for 1 of 2 functionally related genes (endoglin and ALK-1) that

1602 MCELHINNEY ET AL Ann Thorac Surg PULMONARY ARTERIOVENOUS MALFORMATIONS 2005;80:1597 603 encode transforming growth factor (TGF)- receptor subunits involved in endothelial TGF- signaling, which has been implicated in AV differentiation and angiogenic homeostasis [18]. Of note, most of the genes involved in left-right axis determination, and implicated in human heterotaxy syndromes, also encode mediators of TGF- family signal transduction [26, 27]. Patients with heterotaxy, like those with HHTA, may be predisposed to form PAVMs by virtue of disruptions of TGF- signaling, although this hypothesis has yet to be tested. Several investigators [8, 28, 29] have developed animal models to study pulmonary vascular changes, particularly PAVM formation, in the setting of CPA. Studies from these labs have characterized some of the cellular and molecular effects of CPA, and will hopefully lead to an improved understanding of the causes of PAVMs in this setting. Conclusions Hypoxemia resolved after cavopulmonary incorporation of the HV in the majority of our patients with polysplenia syndrome and PAVMs after a Kawashima procedure, presumably due to a combination of resolution of PAVMs and elimination of hepatic venoatrial right-to-left shunting. In a subset of patients with the most profound hypoxemia at the time of PAVM diagnosis and either anatomically advanced PAVMs or postoperative streaming of HV flow to a single lung, PAVMs persisted, typically in the lung that received little or no HV effluent. Of note, creation of an ipsilateral brachial AV fistula in the 2 patients with unilateral HV streaming and persistent PAVMs in the contralateral lung did not result in resolution of PAVMs, in contrast to our experience in patients with unidirectional CPA [30]. In patients with HV streaming, no HV blood reaches the systemic arterial circulation (and hence the brachial AV fistula) without first passing through the lung without PAVMs. These observations support the hypothesis that development of PAVMs is facilitated by exclusion from the pulmonary circulation of a hepatically produced-modified factor that is inactivated-consumed during passage across any capillary bed, and that must be delivered to the affected lung in order for PAVMs to regress. Although the duration of time with a Kawashima circulation did not predict resolution of hypoxemia after HV inclusion, patients with more profound hypoxemia were less likely to benefit from cavopulmonary incorporation of HV flow. Accordingly, we recommend a high index of suspicion for PAVMs, and Fontan completion within 1 to 2 years, in patients with polysplenia and an interrupted IVC who have undergone a Kawashima procedure. Fenestration of the HV pathway may prevent sufficient HV flow from reaching the pulmonary circulation, and should be limited to patients with a specific indication. Alternatively, total cavopulmonary connection without intermediate BCPA may decrease the likelihood of PAVMs in selected patients with balanced pulmonary blood flow. Also, 2 of 4 patients with persistent hypoxemia and PAVMs after HV inclusion had unilateral streaming of HV-PA flow. Thus, approaches that are less likely to be complicated by streaming [22] may be advisable, especially in patients in whom significant SVC-HV offset is likely with unilateral intraatrial or extracardiac redirection of HV return. In patients with unilateral HV streaming, an upper extremity systemic AV fistula is ineffective for treating PAVMs, and should not be performed. Large PAVMs may not resolve after inclusion of hepatic flow, and should be embolized at catheterization prior to HV inclusion. References 1. Bernstein HS, Brook MM, Silverman NH, Bristow J. Development of pulmonary arteriovenous fistulae in children after cavopulmonary shunt. Circulation 1995;92(Suppl 2):II309 14. 2. Chang RK, Alejos JC, Atkinson D, et al. Bubble contrast echocardiography in detecting pulmonary arteriovenous shunting in children with univentricular heart after cavopulmonary anastomosis. J Am Coll Cardiol 1999;33:2052 8. 3. Feinstein JA, Moore P, Rosenthal DN, Puchalski M, Brook MM. Comparison of contrast echocardiography versus cardiac catheterization for detection of pulmonary arteriovenous malformations. Am J Cardiol 2002;89:281 5. 4. Kim SJ, Bae EJ, Cho DJ, et al. Development of pulmonary arteriovenous fistulas after bidirectional cavopulmonary shunt. Ann Thorac Surg 2000;70:1918 22. 5. Vettukattil JJ, Slavik Z, Lamb RK, et al. Intrapulmonary arteriovenous shunting may be a universal phenomenon in patients with the superior cavopulmonary anastomosis: a radionuclide study. Heart 2000;83:425 8. 6. Cloutier A, Ash JM, Smallhorn JF, et al. Abnormal distribution of pulmonary blood flow after the Glenn shunt or Fontan procedure: risk of development of arteriovenous fistulae. Circulation 1985;72:471 9. 7. Kopf GS, Laks H, Stansel HC, Hellenbrandt WE, Kleinman CS, Talner NS. Thirty-year follow-up of superior vena cavapulmonary artery (Glenn) shunts. J Thorac Cardiovasc Surg 1990;100:662 71. 8. Duncan BW, Desai S. Pulmonary arteriovenous malformations after cavopulmonary anastomosis. Ann Thorac Surg 2003;76:1759 66. 9. Kawashima Y, Kitamura S, Matsuda H, Shimazaki Y, Nakano S, Hirose H. Total cavopulmonary shunt operation in complex cardiac anomalies: a new operation. J Thorac Cardiovasc Surg 1984;87:74 81. 10. Srivastava D, Preminger T, Lock JE, et al. Hepatic venous blood in the development of pulmonary arteriovenous malformations in congenital heart disease. Circulation 1995;92: 1217 22. 11. Shah MJ, Rychick J, Fogel MA, Murphy JD, Jacobs ML. Pulmonary arteriovenous malformations after superior cavopulmonary connection: resolution after inclusion of hepatic veins in the pulmonary circulation. Ann Thorac Surg 1997;63:960 3. 12. McFaul RC, Tajik AJ, Mair DD, Danielson GK, Seward JB. Development of pulmonary arteriovenous shunt after superior vena cava-right pulmonary artery (Glenn) anastomosis: report of four cases. Circulation 1977;55:212 6. 13. Uemura H, Yagihara T, Hattori R, Kawahira Y, Tsukano S, Watanabe K. Redirection of hepatic venous drainage after total cavopulmonary shunt in left isomerism. Ann Thorac Surg 1999;68:1731 5. 14. Kawata H, Kishimoto H, Ikawa S, et al. Pulmonary and systemic arteriovenous fistulas in patients with left isomerism. Cardiol Young 1998;8:290 4. 15. Bernstein HS, Ursell PC, Brook MM, Hanley FL, Silverman NH, Bristow J. Fulminant development of pulmonary arteriovenous fistulas in an infant after total cavopulmonary shunt. Pediatr Cardiol 1996;17:46 50.

Ann Thorac Surg MCELHINNEY ET AL 2005;80:1597 603 PULMONARY ARTERIOVENOUS MALFORMATIONS 16. Pandurangi UM, Shah MJ, Murali R, Cherian KM. Rapid onset of pulmonary arteriovenous malformations after cavopulmonary anastomosis. Ann Thorac Surg 1999;68:237 9. 17. Hoeper MM, Krowka MJ, Strassburg CP. Portopulmonary hypertension and hepatopulmonary syndrome. Lancet 2004; 363:1461 8. 18. van den Driesche S, Mummery CL, Westermann CJ. Hereditary hemorrhagic telangiectasia: an update on transforming growth factor beta signaling in vasculogenesis and angiogenesis. Cardiovasc Res 2003;58:20 31. 19. Agnoletti G, Borghi A, Annecchino FP, Crupi G. Regression of pulmonary fistulas in congenital heart disease after redirection of hepatic venous flow to the lungs. Ann Thorac Surg 2001;72:909 11. 20. Baskett RJ, Ross DB, Warren AE, Sharratt GP, Murphy DA. Hepatic vein to the azygous vein anastomosis for pulmonary arteriovenous fistulae. Ann Thorac Surg 1999;68:232 3. 21. Shinohara T, Yokoyama T. Pulmonary arteriovenous malformation in patients with total cavopulmonary shunt: what role does lack of hepatic venous blood flow to the lungs play? Pediatr Cardiol 2001;22:343 6. 22. Lamour JM, Hsu DT, Kichuk MR, Galantowicz ME, Quaegebeur JM, Addonizio LJ. Regression of pulmonary arteriovenous malformations following heart transplantation. Pediatr Transplant 2000;4:280 4. 23. Fewtrell MS, Noble-Jamieson G, Revell S, et al. Intrapulmonary shunting in the biliary atresia/polysplenia syndrome: 1603 reversal after liver transplantation. Arch Dis Child 1994;70:501 4. 24. Gross GJ, Jonas RA, Castaneda AR, Hanley FL, Mayer JE, Bridges ND. Maturational and hemodynamic factors predictive of increased cyanosis after bidirectional cavopulmonary anastomosis. Am J Cardiol 1994;74:705 9. 25. Walker PG, Howe TT, Davies RL, Fisher J, Watterson KG. Distribution of hepatic venous blood in the total cavopulmonary connection: an in vitro study. Eur J Cardiothorac Surg 2000;17:658 65. 26. Bisgrove BW, Morelli SH, Yost HJ. Genetics of human laterality disorders: insights from vertebrate model systems. Annu Rev Genomics Hum Genet 2003;4:1 32. 27. Belmont JW, Mohapatra B, Towbin JA, Ware SM. Molecular genetics of heterotaxy syndromes. 1Curr Opin Cardiol 2004; 19:216 20. 28. Malhotra SP, Reddy VM, Thelitz S, et al. Cavopulmonary anastomosis induces pulmonary expression of the angiotensin II receptor family. J Thorac Cardiovasc Surg 2002;123: 655 60. 29. Starnes SL, Duncan BW, Fraga CH, et al. Rat model of pulmonary arteriovenous malformations after right superior cavopulmonary anastomosis. Am J Physiol Heart Circ Physiol 2002;283:H2151 6. 30. McElhinney DB, Marshall AC, Lang P, Lock JE, Mayer JE. Creation of a brachial arteriovenous fistula for treatment of pulmonary arteriovenous malformations after cavopulmonary anastomosis. Ann Thorac Surg 2005;80:1604 9. Online Discussion Forum Each month, we select an article from the The Annals of Thoracic Surgery for discussion within the Surgeon s Forum of the CTSNet Discussion Forum Section. The articles chosen rotate among the six dilemma topics covered under the Surgeon s Forum, which include: General Thoracic Surgery, Adult Cardiac Surgery, Pediatric Cardiac Surgery, Cardiac Transplantation, Lung Transplantation, and Aortic and Vascular Surgery. Once the article selected for discussion is published in the online version of The Annals, we will post a notice on the CTSNet home page (http://www.ctsnet.org) with a FREE LINK to the full-text article. Readers wishing to comment can post their own commentary in the discussion forum for that article, which will be informally moderated by The Annals Internet Editor. We encourage all surgeons to participate in this interesting exchange and to avail themselves of the other valuable features of the CTSNet Discussion Forum and Web site. For November, the article chosen for discussion under the Pediatric Cardiac Dilemma Section of the Discussion forum is: Outcomes After the Stage I Reconstruction Comparing the Right Ventricular to Pulmonary Artery Conduit With the Modified Blalock Taussig Shunt Sarah Tabbutt, MD, PhD, Troy E. Dominguez, MD, Chitra Ravishankar, MD, Bradley S. Marino, MD, Peter J. Gruber, MD, PhD, Gil Wernovsky, MD, J. William Gaynor, MD, Susan C. Nicolson, MD, and Thomas L. Spray, MD Tom R. Karl, MD The Annals Internet Editor UCSF Children s Hospital Pediatric Cardiac Surgical Unit 505 Parnassus Ave, Room S-549 San Francisco, CA 94143-0118 Phone: (415) 476-3501 Fax: (212) 202-3622 e-mail: karlt@surgery.ucsf.edu 2005 by The Society of Thoracic Surgeons Ann Thorac Surg 2005;80:1603 0003-4975/05/$30.00 Published by Elsevier Inc