Predicting the Identity of Decompressing Veins AfterCavopulmonary Anastomoses

Pamela S. Ro, MD, Paul M. Weinberg, MD, Jonas Delrosario, MD, andJonathan J. Rome, MD

Decompressing veins are an important cause ofunanticipated cyanosis after cavopulmonary

anastomosis (CPA).1–3 Venous connections betweenthe superior and inferior vena caval pathways result incyanosis after bidirectional CPA (hemi-Fontan andbidirectional Glenn), whereas connections betweenthe systemic and pulmonary venous pathways lead tocyanosis after either bidirectional CPA or total cavo-pulmonary connection (Fontan operation).4–6 Previ-ous studies have focused on hemodynamic settings inwhich decompressing channels are likely to occurafter bidirectional CPA.1,2 However the spectrum ofanatomic variations resulting in significant venous de-compression after total as well as superior cavopul-monary connections has not been well characterized.In this study, we defined the anatomy of significantdecompressing veins occurring after CPA, determinedthe incidence of significant decompressing veins byreviewing angiograms in all patients after CPA, anddetermined whether cardiac anatomy or clinical pa-rameters were predictive of decompressing venousanatomy.

The cardiology database from The Children’s Hos-pital of Philadelphia was reviewed to identify patientswith CPAs who had undergone cardiac catheterizationbetween January 1987 and January 1998. Patientswhose angiograms could be obtained and were ofadequate quality for interpretation were included inthe analysis. Echocardiographic diagnoses were re-viewed. The anatomy of the underlying heart defectwas classified into 1 of 4 categories: (1) hypoplasticleft heart syndrome, (2) other functional single rightventricle (including double-outlet right ventricle), (3)functional single left ventricle (including tricuspidatresia and pulmonary atresia), and (4) heterotaxysyndrome. Hemodynamic information including arte-rial oxygen saturation, pulmonary artery pressures,and ventricular end-diastolic pressures was recorded.Angiograms were reviewed noting anatomy and sizeof decompressing veins. Magnification correction wasbased on the known diameter of the angiographiccatheter.

Decompressing vessels were deemed significant ifthey were �3 mm in diameter or associated witharterial oxygen saturation �1 SD below the mean foreach group (superior or total CPA). In many patients,

right azygos veins were intentionally left patent by thesurgeon at the time of bidirectional CPA. The signif-icance of decompressing right azygos veins was de-termined solely by saturation criteria. Finally, in pa-tients with other sources of arterial desaturation aftertotal cavopulmonary connection (baffle leak, fenestra-tion, or partial hepatic vein exclusion), only size wasconsidered in determining the significance of decom-pressing veins.

Differences between groups were evaluated byStudent’s t test or Wilcoxon rank-sum test for numer-ical data, and chi-square distribution for categoricalvariables. A p value �0.05 was considered statisti-cally significant.

Five hundred fifty patients were identified withCPA. Angiograms adequate for review were obtainedfor 298 of these patients. Hypoplastic left heart syn-drome was present in 170 patients (47%), single leftventricle in 101 (28%), single right ventricle in 60(17%), and heterotaxy syndrome in 32 (8%).

Data were collected on 208 patients who had un-dergone hemi-Fontan or bidirectional Glenn proce-dures. The median age at catheterization was 19months (range 4 months to 17 years). Thirty-three ofthese patients (16%) had significant decompressingveins. Pulmonary artery pressures were higher, and(partially by definition) arterial oxygen saturationswere lower in patients with decompressing veins (p�0.05; Table 1).

Four types of decompressing veins occurred afterbidirectional CPA. Most common were azygos con-nections; 19 cases accounted for 58% of the total. Thistype included azygos veins to either the right or leftsuperior vena cava (SVC), hemiazygos veins, andaccessory hemiazygos veins (Figure 1A). Left SVCdecompressing to a coronary sinus occurred in 8 pa-

From the Department of Pediatrics, University of Pennsylvania School ofMedicine and Division of Cardiology, The Children’s Hospital ofPhiladelphia, Philadelphia, Pennsylvania. Dr. Rome’s address is: Divi-sion of Cardiology, The Children’s Hospital of Philadelphia, 34thStreet and Civic Center Boulevard, Philadelphia, Pennsylvania19104. E-mail: rome@email.chop.edu. Manuscript received February19, 2001; revised manuscript received and accepted July 23, 2001.

TABLE 1 Hemodynamic Data from Patients With BidirectionalCavopulmonary Anastomosis

Total(n � 208)

Decompressing Vessel

0(n � 175)

�(n � 33)

Arterialsaturation (%)*

81.2 � 8.6 82.9 � 5.7 72.8 � 14.4

Pulmonary arterypressure (mm Hg)*

12.0 � 3.9 11.7 � 3.7 13.5 � 4.5

Ventricularend-diastolicpressure (mm Hg)

8.2 � 3.2 8.2 � 2.7 8.3 � 5.2

*Significant difference between those with and without decompressing ves-sel (p �0.05).

1317©2001 by Excerpta Medica, Inc. All rights reserved. 0002-9149/01/$–see front matterThe American Journal of Cardiology Vol. 88 December 1, 2001 PII S0002-9149(01)02099-9

tients (24%). In 5 patients (15%), venous decompres-sion occurred as a result of anomalous connectionsbetween systemic and pulmonary veins. Multiplesmall venous connections were responsible for signif-icant cyanosis in 1 patient. The only physiologic fea-tures that correlated with the anatomy of decompress-ing vessels after superior CPA were higher arterialoxygen saturations (82 � 5% vs 70 � 15%, p �0.05)and lower pulmonary artery pressures (10 � 4 vs 15 �4 mm Hg, p �0.05) in patients with decompressing

left SVC compared with other types of decompressingveins.

Data were reviewed from 155 patients who hadundergone Fontan operations. The median age at cath-eterization was 4 years (range 7 months to 31 years).In 22 patients, baffle fenestration had been performed,16 had baffle leaks on angiography, and 15 had un-dergone partial hepatic vein exclusion procedures.Significant decompressing vessels were found in 17patients (11%). Other than lower arterial oxygen sat-uration, neither hemodynamic characteristics nor un-derlying diagnosis was associated with the presence ofdecompressing veins after Fontan (Table 2).

Anatomy of decompressing veins after Fontan fellinto the same 4 groups as after bidirectional CPA. LeftSVC to coronary sinus was most common (n � 7;41%). Flow through a left azygos vein entering the leftSVC below the level of previous ligation (or coilocclusion) in patients with bilateral SVC was presentin 3 patients (18%; Figure 1B). Systemic-to-pulmo-nary venous connections were found in 2 patients(12%). In 5 patients, other types of decompressingvessels were found: 3 had coronary veins that allowedflow from the systemic venous pathway to the pulmo-nary venous atrium, and 2 had hepatic veins drainingto the pulmonary venous atrium. There were no sig-nificant associations between hemodynamic variablesand the type of decompressing vessel.

Channels directly connecting systemic and pulmo-nary veins were responsible for cyanosis in 14% of allcases (n � 7). These vessels originated from the (rightor left) SVC or innominate vein and connected toeither upper pulmonary vein (Figure 2). The unusualnature and unexpected frequency of these vesselsprompted further investigation of these cases. New-born echocardiograms were available for review in 6of these patients. In 5 of 6, there was evidence of leftatrial outlet obstruction at birth. Four patients hadhypoplastic left heart syndrome: 2 with intact atrialseptum and 2 with restrictive atrial septal defect. Thefifth patient had transposition of the great arteries

FIGURE 1. Angiograms representing 2 types of decompressingveins involving the azygos system. A, angiogram obtained froma 17-month-old girl who had undergone a hemi-Fontan proce-dure. The left accessory hemiazygos vein fills after left arm injec-tion. B, angiogram obtained from a 3-year-old girl who had un-dergone completion of the Fontan procedure. The left SVC hadbeen previously ligated. The left azygos vein provides antero-grade flow into the left SVC below the site of the previous liga-tion.

TABLE 2 Hemodynamic Data from Patients With Fontan

VariableTotal

(n � 155)

IntactFontan*

(n � 102)

Decompressing Vessel

0(n � 138)

�(n � 17)

Arterialsaturation(%)†

89.2 � 7.7 90.2 � 7.4 89.8 � 7.8 85.2 � 6.6

Pulmonaryarterypressure(mm Hg)

12.1 � 4.2 12.3 � 4.5 12.3 � 4.3 10.9 � 3.4

Ventricularend-diastolicpressure(mm Hg)

8.4 � 3.9 8.3 � 4.0 8.4 � 4.0 8.5 � 3.9

*Intact Fontan � patients without fenestrations, visible baffle leaks, or partialhepatic vein exclusion.

†Significant difference between those with and without decompressing ves-sel (p �0.05).

1318 THE AMERICAN JOURNAL OF CARDIOLOGY� VOL. 88 DECEMBER 1, 2001

(S,L,L) with left atrioventricular valve atresia and arestrictive atrial septal defect. The remaining patienthad heterotaxy syndrome and total anomalous pulmo-nary venous connection.

Fourteen patients underwent catheter closure ofdecompressing vessels via coil embolization. Thirteenprocedures (93%) resulted in total or subtotal occlu-sion; 1 was unsuccessful. The median arterial oxygensaturation increased 9% with occlusion (0% to 22%)and the median pulmonary artery pressure increased 4mm Hg (0 to 5 mm Hg).

In this analysis, we sought to determine the inci-dence and anatomy of significant decompressing ve-nous channels occurring after CPA. Decompressingveins were found in 14% of patients who had under-gone previous CPA: 16% after bidirectional CPA and11% after Fontan. Four anatomic variants accountedfor �90% of these connections. Most common wereveins of the azygos system, accounting for 40% of thetotal number. These were followed in frequency byleft SVC and veins connecting either the SVC orbrachiocephalic vein to one of the pulmonary veins.Decompression via multiple venous channels wasleast common. No anatomic or physiologic character-istics were particularly useful as predictors for eitherthe presence or type of decompressing veins. How-ever, some associations were found. After bidirec-tional CPA, patients with venous decompression via aleft SVC had higher arterial oxygen saturations andlower pulmonary artery pressures than those withother types of decompressing vessels of which azygoschannels make up the majority. Patients with anydecompressing vein after bidirectional CPA, but notafter Fontan, had higher pulmonary artery pressuresthan those without decompressing veins. This associ-ation between elevation in pulmonary artery pressureand venous decompression after bidirectional CPA isconsistent with previous studies.1,2

We found a lower incidence of decompressingveins after bidirectional CPA than has been previouslyreported (�30%).1,2 This is probably due, at least in

part, to the fact that we chose to limit our study toveins deemed clinically significant. We chose thisapproach out of a concern that virtually all patientswho have undergone bidirectional CPA have venouschannels connecting the superior and inferior venaecavae. Whether such channels are identified may de-pend on how vigorously they are sought out at an-giography. The younger age of our patients comparedwith those in previous reports may also be a factor inthe lower incidence of decompressing veins that wefound after bidirectional CPA. The current data rein-force previous reports emphasizing the importance ofazygos veins as decompressing channels after bidirec-tional CPA.1,2 In contrast, pericardial and mediastinalveins were rarely identified as significant sources ofdecompression in our series. We suggest that althoughsuch channels may often be identified, they are lessfrequently clinically important.

Both interventional cardiologists and surgeons willbenefit from recognition of major decompressing ves-sel categories. Awareness of certain vessels, such aspatent left SVCs and/or azygos veins, will help iden-tify suitable sites for ligation or coil occlusion. Phy-sicians will have a higher index of suspicion for iden-tifying persistent connections between systemic andpulmonary veins in cyanotic patients with evidence ofleft atrial outlet obstruction at birth. These connec-tions are likely present at birth but may not always beidentified on newborn investigations.

The data reported here must be considered in lightof certain limitations. Our results may be biased by thefact that angiograms were either unavailable or inad-equate for review in a substantial percentage of pa-tients who had undergone cavopulmonary connec-tions. Accuracy of the vessel diameter measurementwas limited by the method of calibration. We set outto evaluate clinically significant decompressing veinsand thus defined what in our judgment were reason-able criteria for “significant” veins based on arterialoxygen saturation and size. This definition may not beuniversally applicable.

FIGURE 2. Angiograms demonstrat-ing decompression secondary to sys-temic-to-pulmonary veins (SV-PV).Left, angiogram from a 21-month-old boy who had undergone ahemi-Fontan procedure. A vessel canbe seen connecting the innominatevein to the right upper pulmonaryvein. Right, angiogram from a 12-year-old boy who had undergonecompletion of the Fontan procedure.A decompressing vein connects theSVC to the right upper pulmonaryvein (RUPV). RA � right atrium.

BRIEF REPORTS 1319

Nevertheless, the findings identified in this re-port are noteworthy. Knowledge of the anatomicsubtypes of significant decompressing vessels andthe clinical situations with which they are associ-ated will hopefully aid in the prevention of orexpedient identification and treatment of symp-tomatic decompressing vessels occurring aftercavopulmonary connections.

1. Magee AG, McCrindle BW, Mawson J, Benson LN, Williams WG, FreedomRM. Systemic venous collateral development after the bidirectional cavopulmo-nary anastomosis. Prevalence and predictors. J Am Coll Cardiol 1998;32:502–508.

2. McElhinney DB, Reddy VM, Hanley FL, Moore P. Systemic venous collateralchannels causing desaturation after bidirectional cavopulmonary anastomosis:evaluation and management. J Am Coll Cardiol 1997;30:817–824.3. Gatzoulis MA, Shinebourne EA, Redington AN, Rigby ML, Ho SY, Shore DF.Increasing cyanosis early after cavopulmonary connection caused by abnormalsystemic venous channels. Br Heart J 1995;73:182–186.4. Filippini LH, Ovaert C, Nykanen DG, Freedom RM. Reopening of persistentleft superior caval vein after bidirectional cavopulmonary connections. Heart1998;79:509–512.5. Yoshimura N, Yamaguchi M, Oshima Y, Tei T, Ogawa K. Intrahepaticvenovenous shunting to an accessory hepatic vein after Fontan type operation.Ann Thorac Surg 1999;67:1494–1496.6. Fernandez-Martorell P, Sklansky MS, Lucas VW, Kashani IA, Cocalis MW,Jamieson SW, Rothman A. Accessory hepatic vein to pulmonary venous atriumas a cause of cyanosis after the Fontan operation. Am J Cardiol 1996;77:1386–1387.

Effectiveness of Shielding for Patients During CardiacCatheterization or Electrophysiologic Testing

Alan H. Kadish, MD, Kenneth A. Mayuga, BA, Zachary Yablon, BS,Andi Schaechter, RN, Jeffrey J. Goldberger, MD, Rod S. Passman, MD,Anne Palmer, RN, Michael Zimmer, PhD, and Charles J. Davidson, MD

S tudies have analyzed the radiation exposure andrisk to patients and operators during cardiac cath-

eterization procedures1–7 and from electrophysiologicprocedures,8–13 but have not evaluated the effects oflead groin shielding for patients. This study deter-mines if groin shielding effectively reduces the radi-ation dose to the patient during cardiac catheterizationand electrophysiologic procedures.

This study was approved by the institutional re-view board and all patients provided written informedconsent. To include procedures with longer fluoros-copy times, patients scheduled for interventional pro-cedures or who were likely candidates for coronary orelectrophysiologic intervention were approached andinformed consent was obtained by the research staff.Patients were randomized beforehand to undergo theprocedure either with or without lead groin shielding.Thirty patients scheduled for catheter ablation wererecruited for the study. In 25, a catheter ablation wasperformed and in 5, only a diagnostic study wasperformed. Our goal was to also recruit 25 patientsundergoing percutaneous coronary intervention. Thecardiac catheterization procedures that most patientsundergo at this hospital are diagnostic, with a possiblesame-day intervention; 52 patients were randomizedand were evaluated for a possible intervention. Ofthese, 25 underwent an intervention, whereas 27 hadonly diagnostic cardiac catheterization. Of the 82 totalpatients in the study, 42 were randomly chosen to

have lead shielding (Figure 1). Demographic charac-teristics of the patient population are listed in Table 1.

Groin shielding was performed using a standard,rectangular, commercially available groin shield thatwas 0.05-cm thick and approximately 1,400 cm2 inarea. Before the procedure, the shield was placed onthe table, underneath the examination cloth, in the areaof the patient’s groin. The dimensions of the shieldextended approximately 5 cm above the pelvic ridgecovering the gonadal region, and extended approxi-mately 10 cm beneath the femoral crease. Because ofvariations in patient size and the use of a single-sizedshield, the exact area covered varied slightly frompatient to patient. Commercially available thermolu-minescent dosimeters (TLDs) were obtained fromLandauer Inc. (Glenwood, Illinois). Each square TLDchip measured 4 � 4 mm with a thickness of about 1mm. Nine TLDs were used to measure radiation dosein patients randomized to the use of the lead shield,and 7 TLDs measured radiation dose in those withoutshielding (Figure 2). In both groups, 4 TLD chipswere placed on the table underneath the patient’smidback (1), upper abdomen (2), lower abdomen (3),and groin (4). Three chips were placed directly on thepatient’s upper abdomen (5), lower abdomen (6), andgroin or on the inner thigh as close to the groin aspossible (7). If a shield was present, TLDs wereplaced above it underneath the patient’s lower abdo-men (8) and groin (9). To analyze radiation dose to thelower abdominal and groin regions, chips 8 and 9 inshielded patients were compared with chips 3 and 4 inunshielded patients. After the procedures, the TLDswere removed and sent to Landauer for analysis ofradiation dose equivalent.

For catheter ablation procedures, a Phillip’s SinglePlane C-Arm Diagnostic OP-9 fluoroscopy unit (Phil-lips Medical, Best, The Netherlands) was used toposition catheters. The x-ray tube was below the table

From the Division of Cardiology and Department of Internal Medicine,Northwestern University, Chicago, Illinois. Dr. Kadish’s address is:Northwestern Memorial Hospital, 251 East Huron Street, Suite 8-536,Chicago, Illinois 60611. E-mail: a-kadish@northwestern.edu. Manu-script received June 15, 2001; revised manuscript received andaccepted August 2, 2001.

1320 ©2001 by Excerpta Medica, Inc. All rights reserved. 0002-9149/01/$–see front matterThe American Journal of Cardiology Vol. 88 December 1, 2001 PII S0002-9149(01)02100-2