Keywords
INTRODUCTION
The surgical procedure conceived and performed by Fontan in 1971continues to create a fascinating and growing field of interest inpediatric cardiology. Initially carried out on patients withtricuspid artresia and subsequently used for various defects with asingle, univentricular atrioventricular connection, it is alsoapplicable to other complex cardiac malformations with 2 ventriclescomplete or incomplete when biventricular surgery is notpossible. Thanks to innovative surgeons, the original technique hasevolved throughout the years and undergone various modificationswith the goal of decreasing surgical mortality and prolonging thelongevity of this special circulatory system. In spite of thesecontributions and possibly because of them, the literature embracesthe generic term «Fontan» to describe the procedureitself as well as variants of it and all surgical modalities thatuse a circulation model characterized by the channeling of thevenous return towards the vasculopulmonary outlet without the helpof an immediate ventricular impulse, a model which indeed could bedescribed as the Fontan principle but also as simple totaluniventricular circulation.1-7
Innovations in the procedure include reducing the time betweenprevious palliative procedures and the actual surgery, theimplantation of an extracardiac conduit with or without bypass,fenestration between the systemic venous reservoir and the atriumthat receives the pulmonary veins, and the ability to configure thedefinitive stepped univentricular model into 2 well-differentiatedsurgical stages. Prominent researchers in comprehensive series havedemonstrated the impact of various factors associated withimmediate and/or late mortality that affect surgical success.Disagreement among such researchers is understandable given thenature of the studies and progressive incorporation of variousmedical-surgical strategies.8-17
The objective of this study is to identify the factors thatdetermine overall, early, and late mortality in a series of 124patients undergoing the Fontan procedure and variations inone-stage surgery over a period of 22 years in the same hospital.Other objectives of the study were to compare our observations withthose described in the literature, to propose methods of reducingadverse events in order to decrease mortality, and, finally, to usethe information as a reference for comparing other surgicalstrategies.
MATERIALS AND METHODS
Between June 1978 and February 2001, 166 modified Fontanprocedures and variants of same were performed in our hospital: 131one-stage and 35 two-stage (31 with bidirectional Glenn shunt and 4with previous unidirectional Glenn shunt). Of the 131 children whounderwent one-stage surgery, 7 were excluded due to the presence ofsevere anomalies (discontinuation of the pulmonary branches [n=2],severe organic mitral insufficiency requiring prosthetic valve[n=2], and anomalous pulmonary venous drain [n=3]). We studied 124patients who had surgery between 1978 and 2000. The indication forthe Fontan procedure was established before surgery in 120patients; in the remaining 4 the procedure was chosen duringsurgery due to inability to successfully perform the plannedprocedure, bidirectional Glenn shunt anastomosis in 2 cases andbiventricular correction in the other 2.
Surgery was performed patients with a mean age of 7.3years±4.7 years (0.06 to 35 years), a median weight of 22kg±11.6 kg (3.5 to 90 kg); the male to female ratio was1:47. There were 294 pre-operative catheterizations registered inthe series with a mean of 2.4 per patient (1 to 5), and another 51were carried out in 40 patients post-operatively. Of the pre-Fontancatheterizations, 252 were diagnostic and 42 were therapeutic(balloon septostomy in 36, knife and balloon in 4, and implant ofpulmonary branch stent in 2), anode 18 of the 51 post-Fontancatheterizations had interventions.
Figure 1 shows the various anatomical substrates and associatedanomalies identified on pre-operative angiography andechocardiography, as well as in the surgical notes where necessary.The left ventricle was the principle chamber in approximately 70%of the patients; the right ventricle or a ventricle ofindeterminate morphology in 12% of patients, and 2 completeventricles shared the support of the circulation system in 18% ofthe cases regardless of ventricle size.
Fig. 1. Anatomical diagnosis. TA indicatestricuspid atresia; SLV, single left ventricle; PA+IS, pulmonaryatresia with intact septum; SRV, right ventricle; SindV, singleindeterminate ventricle; 2 VENT, heart disease with 2 completeventricles (1 hyperplasic); AV, atrioventricular.
The number and types of surgery performed prior to the Fontanprocedure are shown in Figure 2. Of 124 patients, 97 had 157surgical procedures (78%) and only 27 (22%) had not had previoussurgery. The procedures were banding of the pulmonary artery (18),systemic pulmonary shunts (87 Blalock-Taussig [36 left and 51right] and 19 Watterston-Cooley; 29 of the total were bilateral),septectomies of the interatrial central separation byBlalock-Hanlon technique (31), and finally aortoplasty had to beperformed in 2 children to correct a coartation of the aorta in 1case and widening of a restrictive interventricular defect in theother.
Fig. 2. Previous surgeries. BAND indicatespulmonary banding; BTL and BTR, Blalock-Taussig left and right;W-C, Watterston Cooley; B-H, Blalock-Hanlon.
There were 102 atriopulmonary connections performed, of which 69were anastomosis between the right atrial appendage and thepulmonary artery (1 with valve conduit) and 33 were between theceiling of the right atrium and the adjacent portion of thesuperior vena cava with the right pulmonary artery; in 4 of the102a Glenn bidirectional shunt was performed (double in 1 case) aspart of the atriopulmonary connection. Sixteen cavopulmonaryconnections were performed (13 intracardiac and 3 extracardiac); in12 of 16 a Glenn bidirectional shunt (double in 2 cases) wasperformed as part of the procedure, while in the remaining 4 theoriginal superior vena cava connection with the right atrium waspreserved during performance of the cavopulmonary technique.Finally, the Kawashima technique was used in 6 patients, connectingthe superior vena cava to the homolateral pulmonary artery (in 1child associated contralateral Glenn shunt was performed) as theywere all missing the suprahepatic portion of the inferior vena cavaand connection of this vessel with the superior vena cava by meansof the azygos; in 4 of the 6 patients the suprahepatic veins werenot incorporated into pulmonary circulation, while in 2 they wereincorporated by means of an intracardiac Gore-Tex®conduit to the right pulmonary artery. In 17 of the 124 patients, 1or 2 bidirectional Glenn shunts constituted part of the Fontanprocedure or a variation of it.
In 22 of 124 children, a 14 mm to 22 mm diameter (mean 19 mm)conduit was implanted in the distinct connections between thevenous system return and the pulmonary arteries. In 13 of 124 casesthe tricuspid valve was closed, isolating the right atrium from thesingle ventricle. Along with univentricularization surgery anyhomolateral stenosis of the pulmonary branch present was corrected;plasty of the contralateral branch was carried out in only 3 cases,and, finally, in 1 child, stenotic lesions in both branches wererepaired. The pulmonary artery was resected or tied off in 104 of124 patients, concurrently occluding systemic pulmonary shunts in67 of the 74 children in whom these were obviously permeable. In 7of the 124 patients a systemic pulmonary shunt was left as anadditional source of pulmonary flow, either due to technicaldifficulties or in the creation of what were significanthypofunctions; in 3 cases these were closed with coils in apost-operative catheterization. In addition, in 2 children othertechniques were combined (1 mitral valvuloplasty, 1 pacemakerimplantation, and small ligature of the right superior vena cava in1 patient and left superior vena cava in 1 patient).
In this series all the children underwent cardiopulmonary bypass(CPB) for 108 minutes±36 minutes (27 to 230). In 7 childrenthe Fontan procedure was performed without aortic cross-clamp; itwas necessary in 117 patients; median time was ±2 min (9 to127 minutes). Rectal temperature was 26±4.6 °C (16 to36). As shown in Table 1, intervention was necessary in 49patients, 33 patients had 47 surgeries to correct residualhemodynamic lesions or rhythm disorders, and 18 therapeuticcatheterizations were performed in 16 patients. Another 13 childrenunderwent 15 immediate minor operations for hemorrhage (n=6),tamponade (n=4), phrenic paralysis (n=3), and mediastinitis (n=2).Six of these 13 children are included in the list of the 33requiring relevant surgical procedures. Six patients required ≥3procedures, 13 required 2, and the remainder underwent 1re-operation.
Follow-up data was obtained from all patients. In those whosurvived more than 31 days, the information was documented at thelast clinical visit or obtained by telephone contact (with parents,cardiologists, and/or the pediatricians involved) during the last 6months before the end of the study (February 28, 2001).
GETTING VARIABLES
Appendices 1 and 2 detail the 45 variables studied, noting theperiods of observation where applicable and the correspondingabbreviations. The anatomic diagnoses were classified into 6subtypes (Figure 1), while principal ventricular morphology wasqualified as left, right, indeterminate, or with 2 ventricles whenboth chambers were observed to be complete. The systolic and medianpulmonary artery pressure (SPAP and MPAP, respectively) weremeasured in all but 8 patients, in 5 of whom the MPAP was obtainedby measuring the buried pulmonary vein pressure. The ratio ofsystemic flow to pulmonary flow was obtained by the Fick method(n=74), and arteriolar pulmonary resistance (apR) was calculatedwhen MPAP and pulmonary flow were determined concomitantly (n=55).Final ventricular diastolic pressure (FVDP) was noted in 121patients, and aortic oxygen saturation was noted in 100 of 124patients. Given the difficulty in measuring the absolute diameterof pulmonary arteries, particularly on older angiographies, wepreferred to determine the ratio of the caliber of each estimatedat the predivision branch level with the corresponding diaphragmaorta. Stenosis located in the pulmonary branch was consideredpresent when the decrease in the vascular lumen was greater than orequal to 50%; similarly, atrioventricular valve insufficiency wasconsidered present when it was moderate or severe. The diastolicsize of the ventricle and its systolic function were bothclassified in a qualitative manner with dichotomic qualificationfrom observation of angiographic images.
Post-operatively, we recorded the amount of time (in days) inthe intensive care unit, the administration of inotropic orinodilator drugs, connection to a ventilator, amount of timethoracic drains were in place, and total time hospitalized. Otherrisk factors with recognized clinical impact as noted in otherstudies were explored in a dichotomic manner, noting at what pointin time they were recorded: left atresia or atrioventricular (AV)valve stenosis in situs habitual, or right atresia or AV valvestenosis in the context of situs inversus; subaortic stenosis byrestrictive interventricular communication or another type ofobstruction with Doppler gradient or by catheter of ≥20 mm Hg;cardiac arrhythmia with rapid or slow disorders of the cardiacrhythm or absence of sinus rhythm; and pacemaker implant. Thedetection of an atrial short circuit by means of cardiaccatheterization or color Doppler echocardiography post-Fontan (byresidual defect in 21 cases, by re-opening the tricuspid valvewhich drives systemic venous blood in diastole to the principalventricle in 5, or because of both conditions in 1 case), this wastermed fenestration and considered involuntary (n=25) if notintentionally created by the surgeon, or voluntary if decided uponby the surgeon (n=2). Protein-loss enteropathy was proven whenthere was hypoproteinemia with hypoalbumenemia and positive proofin Δ-tripsin deposits; finally, the number and type ofpost-Fontan re-operations were noted in detail, with specialattention paid for statistical purposes to those needed to modifyor optimize the univentricular circulatory system (hemodynamic orelectrical); simple immediate re-interventions were thereforeexcluded. A variable called «criteria» was developedwhich defines the risk factors present in each of the children inthe series, using as a reference the Choussat catalogue along withthe most recently published modifications. More than two-thirds ofthe patients had 1 or more risk factors at the time of Fontansurgery (Table 2).18-20
Statistical analysis
Continuous variables were expressed as mean, standard deviation,and range. The median values of the survivor and deceased subgroupswere compared by the Wilcoxon non-paired data test; when thehemodynamic data pre- and post-Fontan surgery were compared, thepaired Wilcoxon test was used. The Fisher χ2 or binomial testwas used to compare proportions, and appropriate contingency tableswere also created. Different variables were analyzed, such ascontinuous or categorized, categorizing deaths, whetherattributable to Fontan surgery or not, as immediate death (≤30days), late death (≥31 days), and both conjointly were qualifiedas dependent. As certain independent variables could be related ina different way to mortality ≥30 days, the possible influence ofeach variable was analyzed using non-conditional logisticregression, and the corresponding odds ratio (OR) with its 95%confidence intervals. For investigation of determining factors inlate mortality, early deaths were excluded from the analysis; alsoas the more recent cases may not have clinical antecedents orrelevant interventions, different variables were analyzed for thosepatients who had been for ≥5 years. Dummy variables were createdfor the pacemaker implant and the appearance of arrhythmia todiscern the impact of each according to the time period in which itwas recorded (pre-, immediately post-, or later). By means of theCox regression method and its corresponding 95% CI, we determinedthe relative risk of the variables so that only those that couldhave a potential clinical influence on surgical outcome wereselected. A retrospective strategy was used in the construction ofthe multivariable models only with those that would have appeared 1P<.1 in the univariable analysis; first the possibilityof co-linearity between variables was eliminated by applying thecorresponding test. Curves reflecting survival rates and thosepatients free of surgical procedures were constructed by theKaplan-Meier method, the CI being calculated by the Greenwoodmethod; this methodology was also used to create MPAP and FDVP, andvarious cut-off points were established as appropriate. Using thesame observation method, we explored the association of the numberof risk factors present in each patient and the type ofuniventricular surgery carried out with overall mortality. Alllevels of P<5% were considered significant. The PRESTAstatistical packet was the basis for all analytical explorationscarried out.21
RESULTS
Overall mortality
The manner and cause of death in the 49 deceased patients isshown in Table 3. Immediate death (n=29) occurred between theoperating room and 28 days post-operatively, with a mean of3.6±7.6 days, while late death (n=20) varied between 0.12and 13 years, with a median of 2.6±4.5 years. Essentially,in more than 60% of patients the cause of death was related toventricular dysfunction, re-operations to optimize theuniventricular circulatory system, or a consequence of aninappropriate procedure. Sudden death related to previously knownarrhythmia occurred in 20% of patients, while the remaining factorsincluded thromboembolism, protein-loss syndrome, neurologicaldisorder, sepsis, and death of unknown origin.
Overall survival at 30 days, 2, 5, and 20 years following theFontan procedure were 78%, 75%, 66%, and 50%, respectively (Figure3). Follow-up of the 75 survivors varied between 0.41 and 20 years,with a mean of 8.4±4.6 years; different clinical aspects ofthis important subgroup will be analyzed in another paper.
Fig. 3. Global survival curve with Fontansurgery and variants, with corresponding 95% confidence intervals.Notice that like initial mortality, the obit continues toprogressive through the years. The number of patients beingfollowed appears in parenthesis.
Univariate analysis comparing the mean values of continuousvariables between those who died and the survivors is shown inTable 4. The differences in MPAP, FDVP, and the thoracic drainagetime between both subgroups are significant, although the medianvalue is affected by the short course of the thoracic drain implantin those children who died early. Although not of statisticalsignificance, the median values of body weight, apR, and the amountof time in intensive care show a tendency to be lower in the livesubgroup with respect to the dead subgroup.
The date of surgery shows a slight relationship with overallmortality. In effect, 57% (27 of 47) of those undergoing surgery1978 and 1989 died, vs 29% (22 of 77) of thos e who had surgerybetween 1990 and 2000 (χ²=9; P<.003), althoughit is true that there is a strengthening effect on the differencesin mortality attributable to the progressive number of patients whodied during follow-up. Figure 4 shows the survival curve forpatients who had surgery during these 2 observation periods.
Fig. 4. Survival curve according to the surgicalcalendar: patients operated on between 1978 and 1989, and between1990 and 2000. Differences are noted from the immediatepost-operative period.
Table 5 shows the individuals and percentages that correspond tothe various dichotomic variables. The creation of contingencytables for each did not show an association with mortality, exceptwhen evaluating AV valve insufficiency (P<.017). Atendency toward that association was observed when compared withthe presence of ventricular dilatation (P<.081);nevertheless, the qualitative dichotomic methodology used to typifycancels out this observation. Of the 8 children with smallpulmonary branches (right branch in 2 and both branches in 3patients) 70% died compared to 37% of those who died that hadnormal branches (P<.057).
Ventricular morphology was not associated with mortality, eitherwhen the original categories were analyzed or when those childrenwith a left ventricle serving either as the only ventricle or theprincipal ventricle are grouped together compared with right orundetermined ventricular morphology. There were no differencesobserved when the subgroup with 2 ventricles was incorporated intothe analysis. The number and type of previous surgeries had nobearing on overall mortality when analyzed with contingency tables.Figure 5 illustrates survival according to the type of surgicaltechnique used. Comparison of the survival curves shows a tendencyto differences (χ²=4.8; P<.08); but the analysisis influenced by the small number of patients who underwent theKawashima technique, which produces an excessive CI for thecorresponding curve. Even assuming that the atriopulmonarytechnique was used more often and is an older technique than thebicavopulmonary one, the comparison of outcomes between the 2techniques alone is significant when expressed as curves(χ²=4.4; P<.03) as it is when compared viacontingency tables (χ²=6.8; P<.008).
Fig. 5. Survival curve in relation to thesurgical technique used. Although the differences between thedistinct techniques are appreciable, it must be noted that thebicavopulmonary technique is more recent and has had less time forfollow-up.
Table 6 lists the variables associated with overall mortality asevaluated by the Cox regression method. Univariate analysisdemonstrated that MPAP, FDVP, CPB, atresia or mitral stenosis, thecriteria, the combined association with situs inversus andvisceroatrial heterotaxia, and AV insufficiency are narrowlyassociated with mortality; with the exception of AV insufficiency,the other variables remain in the multivariate model.
The impact of MPAP and FDVP on surgical results, expressed assurvival curves, is shown in Figures 6 and 7, respectively. For theanalysis MPAP, the various categories were explored, and it wasfound that 2 showed major differences between them (≤18 vs≥19 mmHg, respectively) (χ²=6.3; P<.011); inthe same way, the cut-off point for statistical significance forFDVP was ≤12 vs ≥13 mm Hg (χ²=8.9; P<.002).
Fig. 6.Survival curve which illustrates theimpact in the global evolution of pre-operative medianarteriopulmonary pressure; patients with ≥19 mm Hg had a worseprognosis.
Fig. 7. Survival curve which illustratesthe global evolution of patients with ≤12 and ≥13.
Figure 8 shows the survival curve for the criteria variables. Itwas determined that 3 categories most faithfully represented theirassociation with overall mortality; children with 3 or moreassociated risk factors had an essentially worse prognosis(χ²=20.4; P<.00006).
Fig. 8. Survival curve which shows thenegative impact on initial and late mortality according to thenumber of pre-operative risk factors present in each patient.
Variables associated with early mortality
The surgical timeline, although without statisticalsignificance, showed a relationship with regard to mortality ≤30days, as it reveals a decrease from 30% in the period of 1978-1989to 19% in the period. In the last 3 years, only 1 death has beenrecorded, which is 8% of early death in this period (1 of 13).
Age at the time of surgery was associated with immediatemortality, as 56% (5 out of 6) of patients younger 3 years of agedied early while only 21% (24 of 115) of patients older than 3years died early (χ²=3,8; P<.47). Body weightwas associated with early death: 47% (8 of 17) of the children whoweighed ≤13 kg died vs 20% (21 of 107) of those with a weight of≥14 kg (2 = 4.6; P<.029). Table 7 shows the otherassociated variables which were analyzed by non-conditionallogistical regression: FDCP, CPB, atresia or mitral stenosis,fenestration, criteria, and situs inversus.
When observed categorically, the impact of CPB and aorticcross-clamping can be appreciated. In effect, 30% (24 of 79) of thechildren with CBP≥91 min died vs 11% (5 of 45) of those with≤90 min (χ²=4.9; P<.02); in the same manner,32% (22 of 68) of children with a clamping time of ≥61 min died,vs 14% (7 of 51) with a time of ≤60 min (χ²=4.5; P<.03). The presence of mitral atresia or severe stenosis ofthe mitral valve was associated with mortality; it was notdetermined that there was an association due to the coexistence ofatrioventricular valve insufficiency, which acts as the only ordominant valve. It is of interest to note that the presence offenestration is associated inversely with early death; only 75 (2of 27) of patients with this diagnosis died, while 28% (27 of 97)of those with no sign of a residual short circuit eitherbecause there was no record of systemic oxygen desaturation orbecause it was excluded by catheterization and/or echocardiography(χ²=3.8; p<.042). The impact of this variable on theimmediate outcome is the same even those children who died duringsurgery were excluded; in these children it was impossible to knowwhether a residual atrial defect was present. This observationmeans that fenestration has a protective effect on the immediatepost-operative outcome.
Other variables (while not reaching statistical systemicpulmonary shunt implant (OR=2.1; 95% CI, 0.88-4.99). With regard tothe latter variable, median FDVP in children with shunts wassignificantly higher than those without shunts (12 vs 9,respectively; P<.012). Although the morphology of theprincipal ventricle did not show a relationship with early death,it must be pointed out that 60% (6 of 10) of the children withsitus inversus and heterotaxia died, vs 20% of those with habitualsitus visceroatrial (23 of 114) (χ²=6,1; P<.01).This difference most likely probably reflects a constellation ofvariables present in children with situs other than the habitualtype which supposes a more serious pathology. In the multivariantmodel, only FDVP, fenestration, and atresia or mitral stenosis werepresent.
Variables associated with late mortality
We analyzed 95 patients who lived ≥31 days. The variablesassociated with late mortality are shown in Table 8, and were MPAP,FDVP, re-operation, criteria, and surgical timeline. We alsodetermined that both the protein-loss syndrome (OR=5.01; 95% CI,0.90-28) and subaortic stenosis (OR=2.17; 95% CI, 0.96-4.91) had anappreciable negative impact on late mortality, even when it did notreach the required statistical significance. In the multivariatemodel, only the criteria variables and re-operation were left.Figure 9 shows the survival curve for patients without post-Fontanre-operations or catheter re-intervention, excluding thoseprocedures not associated with repair of hemodynamic sequelae orcardiac rhythm; there is a progressive need to correct alterationsthroughout follow-up. Only those patients with ≥5 year follow-up(9.7±3.5 years; between 5 and 20) were analyzed in anattempt to avoid bias which assumes the inate pacemaker implant areassociated with mortality at ≥5 years (Table 9).
Fig. 9. Survival curve of patients withoutsurgical or catheter re-intervention during the follow-up period.Note that in the first 12 years the procedures are necessary toimprove or abolish residual lesions.
Other observations of clinical interest
Subaortic stenosis
Ten chindren (8%) were included in this subgroup with a subaortic gradient of between 20 and 100 mm Hg and an overall mortality rate of 40%. In 8 of 10 the obstruction was detected after the Fontan procedure. Banding (60% vs 11%) (χ²=14; P<.0002) andventriculoarterial discordance (100% vs 61%) (χ²=4.4; P<.033) had a significant association with subaorticstenosis with respect to those who did not have the condition. Sixchildren underwent surgical repair: 1 before univentricular surgeryand 5 as re-interventions in the post-Fontan period (between 3 and9 years later); 1 of the latter patients required 2 procedures toalleviate the obstruction and ultimately died 7 years post-Fontan,after being rejected for cardiac transplant because of the numberof prior surgeries. Of those in whom no surgery was considered toalleviate the subaortic obstruction (4/10), 2 children diedpost-Fontan procedure (1 late and 1 immediately); another 2survived, although 1 of them developed protein-loss syndrome and iscurrently in excellent clinical condition after a successfulcardiac transplant performed 6 years post-Fontan procedure.
Protein-loss syndrome
Five children (4%) developed protein-loss syndrome; 2 (40%)died, 1 from hepatitis contracted months after 2 re-operations(repeat Fontan and diaphragmatic kink) and the other died suddenlyof known arrhythmia. In the 3 surviving patients, re-operationswere carried out from 2 to 12 years post-Fontan: 1 closure of aresidual short circuit, 1 cardiac transplant, and 1arteriopulmonary closure. In addition, no pre- or post-operationhemodynamic variable was associated with protein-loss syndrome.
Arrhythmia
This subgroup was made up of 41 children (33%). The rhythmdisorder was detected in the pre-Fontan period in 8 patients(19.5%); in the immediate post-operative period in 12 (29%), andduring the procedure in 21 (51%). Atrial flutter, paroxysmaltachycardia, and advanced AV block were the most frequentarrhythmias found. Thirteen children died (32%), 2 during thepost-operative period of ≤30 days. Twenty-seven percent (11 of41) died during follow-up (7 suddenly), a significantly differentnumber than the 11% late mortality rate in the patients withoutpreviously known arrhythmia (P<.022). In 13 of the 41patients various re-interventions were necessary, 12 involvingpacemaker implantation. Pre-operative MPAP was higher in patientswith arrhythmias as compared to those who did not have arrhythmias(15 vs 13.7; P<.02); when patients with pre-operativerhythm disorders were excluded, the MPAP lost its significance as apossibl e factor in inducing arrhythmia. Previous interatrialseptectomy (Blalock-Hanlon) was also not associated withpost-Fontan arrhythmia; on the other hand, it was observed that theatriopulmonary technique showed a marked tendency to be associatedwith an altered post-operative rhythm with the cavopulmonarytechnique (19 vs 2, respectively; P<.07).
Thromboembolism
The presence of systemic venous thrombosis was evident ontransthoracic echocardiography (TTE) in 5 patients (4%). Earlydemise was noted, preceded by confirmed pulmonary thromboembolism,and 3 patients died later (of jugular vein thrombosis and superiorvena cava thrombosis), all within the context of re-operations tooptimize circulation. All the patients who died had undergone theatriopulmonary technique. In 1 survivor an inferior vena cavathrombosis was detected immediately post-operatively, whichdisappeared with heparin treatment. Thromboembolism was suspected,but not definitively confirmed, in 2 children who died in the earlyperiod (probably of pulmonary embolism) and later (probably fromsystemic embolism in the presence of fenestration).
DISCUSSION
Our series and its limitations
The current study analyzes and describes in detail ourexperience in a large series of children operated on with one-stageFontan surgery or a variant thereof performed at the same centerover a period of 22 years. The study population consisted primarilyof cases similar to those in the first large studies published inthe literature, which involved patients operated on in the80´s and in which the predominant technique of choice wasatriopulmonary connection. Our series some years laterextended well into the 90´s; only in the later yearsfrom 1992 on have we gradually included thebicavopulmonary technique.22-28
Given the changes in strategy for patient selection and choiceof appropriate surgical technique over the years, and because it isretrospective, the current study has design limitations which mustbe considered when interpreting the information detailed in thediscussion. Given the well known hemodynamic considerations,therefore, we have chosen not to include information gleaned fromtwo-stage surgical univentricularization, although this last groupof patients will be documented in a future paper.
Because of ethical considerations and limitations of thesurgery, it is essentially impossible to randomize homogenoussubgroups which allow the clear perception of factors thatcontribute to surgical failures and the lack of durability of thiscirculatory system. Even with these precedents, we believe thisstudy is broad for a single center and that it yields relevantinteresting data for both the clinician and the surgeon that effectallows the Fontan principle to be examined from differentperspectives. The study also serves to inform is as it highlightsthe results of new strategies that we are currently applying.
Mortality
Overall mortality in our series was 39% (49 of 124 patients) ina follow-up of up to 20 years (8.4±4.6), with survival of 30days (78%), 2 years (75%), 5 years (66%), and 20 years (50%)post-surgery (Figure 3). At the Mayo Clinic, of 352 Fontansurgeries between 1973 and 1984, 16% died within a month, 23%within a year, and 40% at 10 years; in a multicenter study of 334children operated on during the 1980s, 33% died early and 13%later.24,29
Our overall mortality with this one-stage surgery is high,including for those cases that occurred in the 1990s. This may bedue to an initial prodigal attitude regarding palliativetechniques, and to performance of a technique that was notsufficiently established and that had a high incidence ofre-operations, all factors which we know cause deterioration ofventricular function. Since the immediate mortality rate hasdropped to 8% in recent years, 22% (29 of 124) is a high rate; insimilar series carried out in the 1980s and 1990s, between 15% and31% of children died in <30 days. In more recent studies,mortality drops to between 0% and 10%, increasing survival to morethan 80% at 10 years.8,22,30-35
Risk factors related to the patient
Ventricular function and its enemies
Since the ventricle acts as a pump and the univentricular systemis the only energy source to support 2 serial resistances, goodfunction is indispensable for the success, both immediate andlater, of the Fontan operation. In this circulatory model, 90% ofventricular energy is dissipated in the blood flow impulse throughrecurrent systemic vascular resistance in the best cases, and theremaining 10% by its own diastolic suction power overcomespulmonary vascular resistance; therefore any decrease inventricular function, no matter how small, could be critical to aloss of the energy needed for blood flow to easily cross thepulmonary vascular circuit.13,14
Ventricular function in malformations in which there is only 1useful ventricle can deteriorate for many reasons, the primary oneprobably being the substrate of the myocardial myoarchitecture asdescribed by Sánchez Quintana, other reasons being thenatural course of the disease or the surgical modification. Initialpre-operative ventricular blood volume tends to more than doublewhat is considered normal in biventricul ar hea rts. The numeroussys temic pulmonary shunts inevitable in the majorit y ofcases (in our series 106 in 77 of the 124 patients) produce apulmonary hypervascularity which favors good initial oxygensaturation in these patients, but also increases the alreadyexcessive ventricular overload that is present and that often isreinforced with atrioventricular valve(s) insufficiency of varyinggrades.36,37
The chronic volume overload produces a geometry which progressesto dilatation. If the volume to mass ratio does not balance withprogressive compensatory parietal hypertrophy or the dilatation isexcessive, changes in systolic and even diastolic function canresult that affect the ventricular suction power. Although it isdifficult to predict when such changes are irreversible, it seemslogical to correct the volume overload as soon as possible,avoiding shunts if possible or implanting the smallest caliberpossible, or by performing Glen anastomosis or Fontan surgery, orboth, early on. In light of this considering that the lattersurgery induces a rapid geometric change, the relative predominanceof hypertrophy in these conditions can lead to a loss of myocardialelasticity, with resultant diastolic dysfunction during the firstpre-operative days or months. These effects are evinced in study bythe fact that previous shunt implant and ventricular dilation, withits consequent influence on final diastolic pressure, had atendency to be associated with a worse surgicaloutcome.38-44
In the same manner, in anatomical substrates post-operativegeometric change facilitates subaortic obstruction (by reducing thesize of the interventricular communication) or favors AV valveregurgitation because of valve redundancy. At least 20% of patientswith a single ventricle have systemic obstructive lesions thatcause a pressure overload and ventricular hypertrophy that altersdiastolic function. In these patients there is often an associatedanterograde pulmonary hyperflux due to the absence of protectivepulmonary stenosis; the inevitable need to control pulmonary flowwith banding produces or increases subaortic stenosis, changingventricular function even more. In our series, as is well known, wealso observed a narrow relationship between banding,ventriculoarterial discordance, and the appearance of subaorticstenosis.45-50
The preceding effects would result in severe ventriculardysfunction immediately post-surgery, with consequent patientdeath; therefore the concept of the two-stage Fontan principle hasbecome more and more refined so that changes relating to the massto volume ratio have progressed, which we have also noted in recentyears. It remains to be seen if this strategy results in lowersurgical mortality rates and, at the same time, greater longevityof the circulatory system. In any case, since function of the onlyimpulse pump is essential in the physiology of the univentricularmodel; it seems evident that many of the deleterious forces thatact upon it can affect, and also exhaust the limits of its reservefunction. In the majority of studies, as well as in ours,ventricular dysfunction is the principle cause of early and latedeath, from the Fontan dismounting or from cardiac transplant.Given this, it is not surprising that indices of ventricularfunction performance, such as FDVP, ejection fraction, hypertrophygrade, and cardiomegaly are universally recognized as riskfactors.22,22,34,51
In our series FDVP is a determinant of mortality during anystage of the study and the greatest impact on prognosis; so that wecan be certain that an FDVP>12 mm Hg indicates a negativesurgical outcome, as shown in Figure 7, even though the survivalcurves in those patients with ≤12 and ≥13 mm Hg tend to beclose on follow-up. We also observed that ventricular dilatation,which in spite of its dichotic determination can be consideredimprecise, tends to be associated with overall mortality(χ²=2,8; P<.08). In light of this data it isessential to try to preserve the maximum ventricular functionreserve from the neonatal period on; therefore we believe it isnecessary to have a new strategy of early intervention and avoidpalliative surgery as much as possible by trying to controlpulmonary flow when by early implantation of bidirectional Glennanastomosis or Fontan surgery when or variant thereof when it isinevitable and, when necessary, optimize hemodynamics byconfiguring total two-stage univentricularization. The future willreveal whether this assertion is true; we would not hesitate toconsider cardiac transplant if ventricular function hasdeteriorated considerably.12,13,17
Atrial valve insufficiency. Atrial valve insufficiency isa known risk factor for immediate and overall mortality in Fontansurgery; the 5 children with this condition died (Table 5). Itsassociation with the presence of a common or unique AV valve wasevident every time this valve morphology was present in 3 of them,who also had visceroatrial heterotaxia. In a series of 500 casesundergoingFontan surgery, the Mayo Clinic identified AVinsufficiency as an immediate post-surgical risk factor; thisassociation is confirmedthis time with overall mortalityratesin 352 surgical cases between 1973 and 1984. Kaulitz etal, in a study of 72 children with cavopulmonary surgery, alsorecognized AV insufficiency as a significant risk factor. In fact,it is associated with an increased risk of death in visceralheterotaxia syndromes with or without abnormal connections ofpulmonary or system veins, malformations in which only 1 AV valveis frequently present.24,37,52-55
These observations are not surprising considering the fact that1 morphologically normal functionally competent valve is essentialfor good system function. As long ago as 1977 the absence of mitralinsufficiency was requirement number 9 on the list of the 10requirements of Choussat, and even of those at least 5 seem to havecurrently lost importance, this continues to be fundamental in thesuccess of univentricular surgery. It is not unusual to encountercases of a mitral prosthesis in a Fontan series; nevertheless, fewknow the later outcome of these patients. Speaking from our ownexperience, it is recommended that mitral valvuloplasty beperformed before any major insufficiency occurs. Although themethods we used to typify the grade of regurgitation differs fromthat described by Imai et al, the authors made similarrecommendations, and even when discussing the immediate and laterisk of this condition, they recognized a reduction in valvecompetence through annuloplasty in those patients with apre-operative regurgitation grade higher than2.18,37
Ventricular morphology. Although there is not adefinitive consensus on the impact of ventricular morphology onmortality in Fontan surgery, many researchers consider right and/orindeterminate morphology to be prognostic factors in terms of bothmortality and morbidity. The failure of the chamber in question toachieve normal aerobic capacity with exercise in children andadults who undergo surgery is established, supporting the findingsof abnormal diastolic function and deficient functional reserve. Inour study we did not observe this finding, but perhaps this is dueto a potential problem because of the small number of patients withthis type of morphology. On the other hand, we did observe that theconstellation associated conditions related to the AV connection orto visceral situs, among others, is greater in those cases with adominant right or indeterminateventricle.17,30,56-59
Mitral valve stenosis/atresia. Mitral valve stenosis oratresia is a determining factor for mortality in our series; 8 ofthe 10 children (80%) with this anomaly died. Many mechanisms aresuspected in this anomaly such as: a) pulmonaryvenocapillary congestion and passive hypertension over severalmonths and even years, exacerbated by the coexistence of previoushyperflux, which would alter the extrauterine maturation of thepulmonary vasculature (3 old cases in our series); b)balloon and/or knife atrial septostomy, and the need forBlalock-Hanlon septectomy (7 out of 10 of our patients) could bethe cause of additional interventions that could induce arrhythmiasand, in turn, difficult later surgeries and the atriopulmonaryconnection in the presence of mitral stenosis or atresianecessitate oblique and wide patch implantation, with long suturelines that not only increase CPB time but also potentiate theappearance of arrhythmias. Currently, since the cavopulmonarytechniqueintra- or extracardiacis the technique of choiceat most centers, this last mechanism is less important; werecognize that to achieve early atrial communication would softenthe impact of the two first points.48,60,61
Systemic obstructions. Aortic coartation and aorto priorto Fontan surgery have also been identified as overall mortalityrisk factors in a multivariate model for the Boston group. If thisis confirmed by other studies, the physiopathologic mechanism maybe related to parietal hypertrophy of the ventricle, althoughpulmonary hypertension cannot be rules out in cases with hyperflux,which is also common in these children. In our series only 1patient required pre-Fontan aortoplasty, and they survivedsurgery.30
With respect to subaortic stenosis, 4 of our 10 children (40%)with this condition died. Although there was a tendency ofsubaortic stenosis to be associated with late mortality, thenegative impact of the obstruction was very evident when we includein the subgroup the only patient who developed protein-losssyndrome associated with the subaortic obstruction and who survivedcardiac transplant. The Mayo Clinic group established a strongassociation between the subaortic gradient and mortality in Fontansurgery; mortality can reach 67% if the gradient is more than 40 mmHg, or be as low as 17% if the defect is widened first, and is lessthan 10% with a gradient of less than 25 mm Hg. In most of ourpatients this condition appeared post-Fontan surgery; for thisreason it is essential to determine the pre-operative sub-aorticgradient where there is an anatomical tendency, as the latentobstruction could be circumvented at the time of or before Fontansurgery.48,51
Situs inversus and visceroatrial heterotaxiavisceroatrial
Whether individually or in combination, situs other thanhabitual is a known mortality risk factor. In our series 4 of the 6children with heterotaxia died (67%, 3 immediately post-surgery and1 later), and 3 of the 4 with situs inversus (75%, all in≤30 days). In the Mayo Clinic experience, heterotaxia is adeterminant of poor prognosis both in those undergoing theatriopulmonary technique and the cavopulmonary technique.Marcelletti also observed that heterotaxia increased surgical riskin patients with extracardiac conduit surgery. An early Glennstrategy to repair valvular insufficiency, which frequently occursin these cases, could benefit the course of these children; buteven so it must be determined whether this complicated subgroup hasa good long-term prognosis with Fontansurgery.34,54,56,62
Arrhythmias. An arrhythmia can occur pre-, post-surgery,or immediately, it is difficult to clearly determine the impact ofa rhythm disorder on patients who undergo Fontan surgery orvariants there of. In our series, we observed that onlylate-appearing arrhythmia was associated with mortality in childrenwho followed more than 5 years. In fact, 5 of the 7 who died had ahistory of recognized arrhythmia. There is no consensus on whetherrhythm disorders occur less frequently in extracardiac tunnelprocedures with respect to the atriopulmonary technique in thelong-term. In any case different factors can induce arrhythmia. Aswe noted, we also observed an association of pacemaker implant andlate mortality; the Mayo Clinic group found that this is greater inchildren requiring pre-Fontan pacemaker implant, while the Bostongroup noted its impact on children who underwent intracardiactunnel technique and pacemaker implant pre- and post-surgery. Whatdoes seem evident is that the presence of rapid or slow rhythms orthat are congenital or occur post-surgically can affect ventricularfunction and, in turn, lead to frequent re-operations that worsenthe prognosis for these children. Aggressive correction of rhythmdisorders is desirable, and the hemodynamic or electrical situationof each patient must taken into consideration, and the disordermust be treated aggressively with medication or possiblereconversion by the atriopulmonary or cavopulmonary technique alongwith anti-arrhythmia surgery if necessary.62-70
Pulmonary circulation and vascular anatomy: the last barrierto overcome. Arteriolar pulmonary resistance which is dependenton adequate anatomy and physiology of its vascular model mustnecessarily be low for Fontan surgery or its variants to besuccessful. We point out that, in most cases, the ventricularimpulse only has 10% of the available energy to cross the pulmonaryvascular barrier. Therefore, any increase in resistance, no matterhow slight, can have disastrous affects on the fragile Fontanprinciple. There are many factors in the patients´ naturalhistory and the surgical modifications that can increaseresistance; these factors are physiological (long-term pulmonaryhyperflux, obstructive systemic lesions), as well as anatomicalsecondary to vessel distortion, amputation of secondary branches orarterial stenosis resulting from shunts, occlusion of ducts in somecases, or occlusion of the pulmonary artery. All these conditionsmay restrict the capacity of the vascular pulmonary tree, and it isno surprise that these physiological or morphologic changes havebeen defined as risk factors. In our series, MPAP has a clearimpact on the prognosis; when greater than 18 mm Hgpre-operatively, there was a negative survival curve (Figure 7). Wehave proved, furthermore, that with more than 15 mm Hg of median APpressure, the prognosis worsens. There are a number of reasons forthe influence of apR on surgical outcomes such as the errorsinherent in the surgical method in cases where this could bedetermined, and the small number of indications used to perform thesurgery. Two of the 3 cases that exceeded the 2.5 u/m² died(67%). We have also explored other contraindications, such as thosedescribed by Mair and Knott-Craig, related to different hemodynamicvariables in pulmonary and systemic flow, MPAP, and apR. We havenoted its association with immedi ate and overall death but, giventhat its predictive value was not greater than that of MPAP, we didnot report it in this article.62,71,72
In the Boston group an apR >2 u/m², MPAP>15 mm Hg,and the distortion of the pulmonary arteries constituted clear riskfactors. Kirklin also established that this last variable isassociated with overall mortality. In a combined multicenter study,it was recommended that successful surgery required the absence ofdeformities of the pulmonary vascular tree, and Nakata and McGoonindexes greater than 250 mm²/m² and 1.5. Although thesedata have been questioned by other researchers who argue that thecontinuous non-phasic flow characteristic of this univentricularcirculatory model makes it unnecessary to attain those values. Inour series, when both branches are small (≤.75 as related to theabdominal artery), the surgical result is worse (3 patients withthis condition died). The influence of stenosis of the branches onthe outcome was not evident. Repair of same during surgery withoutexcessive CPB prolongation or later catheter interventionneutralizes its potentially negativeeffect.18,24,30,34,62,73-75
Other extrapulmonary and extracardiac factors
Younger age (<4 years) and lower body weight (<15 kg)tended to increase the risk in surgical treatment of congenitalheart disease. There are many reasons for this such as the greatercomplexity of heart disease, the smaller margin of error in thetechnique, or difficulties in post-operative management, amongothers. Although several researchers have also highlighted theimportance of age and weight as determinants of mortality inuniventricular palliation (both in atriopulmonary surgery and withplacement of conduits), most surgeons believe that technologicaladvances and improved techniques cancel out the negative impact ofthese factors. In fact, in a series from the 1980s and in morerecent series it was determined that age was not a risk factor formortality; it was determined that Fontan surgery could be performedsafely at 3 years or younger. Given that our group had successfullyundergone cavopulmonary surgeryin 1 or 2 stagesand basedon our own observations, we established an age of about 3 years anda body weight of about 13-15 kg asappropriate.28,76,77
The variable criteria consisted of 10 cardiac, pulmonary, andage factors. In our series, the 10 factors that reflectedventricular function, pulmonary vascular situation, and age had astrong impact on early, late, and overall mortality, as can be seenin the corresponding tables. When survival is compared with thenumber of factors present in each child, it is evident thatchildren with 3 or more criteria had a worse survival curve (Figure8). Other authors have developed similar significant criteria thatare useful in determining risk for futurepatients.24
The protein-loss syndrome is an undesirable complication of thesurvivors of Fontan surgery. The rate of 4% in our series issimilar to that described in a multicenter study by Mertens et al.This condition has a poor prognosis, and its actual mechanism isnot well understood, although it is accepted t although mediancavopulmonary pressure does not tend to be elevated in these cases,as in our experience. The success of re-operations to repairsignificant residual lesions in these children is lower in thestudy of Mertens et al; however, we believe it is necessary totreat these lesions surgically or with catheterization, as we wereable to do in 2 of our patients. Cardiac transplant is also anoption when re-intervention to correct functional anatomicalterations is not possible, or when the clinical picture does notchange despite recent heparin treatment (1 child in ourseries).78-80
Thromboembolism is another condition that increases mortality inFontan surgery. The observation in our own patients of a very slowflow parallel to the wall of a dilated wall and a hypocontractileright atrium on post-surgical angiography or evidence ofsignificant autocontrast on echocardiography in this chamber and inthe venas cavas are a result of changes in multiple thrombogenicfactors (anatomofunctional and biochemical) that pose athromboembolic risk. We were able to identify early signs ofthrombosis using transthoracic and transesophagealechocardiography, and aggressively treating this hematologicalchange.81
Factors determining mortality depending on time andtechnological progress
Date of surgery
As is proven in our series, the surgical timeline has a strongimpact on immediate, late, and overall mortality. We observed notonly the loss of patients over the years, but also a reduction inimmediate mortality to 8% over the last 3 years. The year surgeryis performed is also an omnipresent factor in all the large seriesthat span many years, and this is attributable to the learningcurve, the advent of ne w technologies, and the disparate coursethat results from differe n t surgical techniques. As a result, thedate of surgery is universally considered to be a determiningfactor in mortality. This finding is not new, but it confounds thedetection of other contributing factors of greater importance.
Fontan technique
As this is the oldest, the atriopulmonary technique is adeterminant of mortality; the overall rate in our series of 102cases was 44% and only 6.3% in those cases who underwentcavopulmonary surgery. Figure 5 shows the lowest survival rate inthose children undergoing on the Kawashima and atriopulmonarytechniques. This study, however, is not optimal for studying thesetechniques, as it does not take into account the tenuous time offollow-up. Also, as we have shown in our own series, patientselection has improved drastically over the years: the percentageof children without risk factors has increased from 29% with theatriopulmonary technique to 68% with the bicavopulmonary technique(P<.04). Balaji et al also observed that thecavopulmonary technique reduces overall mortality from 37% to 15%,and they point out that the incidence of arrhythmias is lower withrespect to established surgical techniques. Although variousauthors have also noted the advantages of the cavopulmonary model,others have not documented such differences. These discrepancies,therefore must be attributed to the difficulty of carrying outcomparative study series simultaneously and with the samefollow-up, although the conventional wisdom is that thecavopulmonary technique has appreciableadvantages.32-35,51,52,63,82
Cardiopulmonary bypass time
As in many other series, we found that prolonged CPB timeaffected early and overall mortality rates. When CPB was >90min, mortality increased from 34% to 42%. This was also apparentwith aortic cross-clamping (31% vs 44% when >60 min). Althoughboth variables are subject to multiple intraoperative variables,their deleterious effects on endothelial myocardial and pulmonaryfunction are established, especially the damage on univentricularcirculation since this depends on good diastolic and ventricularfunction and low pulmonary resistance. We and other researchersbelieve that aortic cross-clamping and even cardiopulmonary bypassmust be avoided whenever possible in Fontan surgery. Although wecould only avoid aortic cross-clamping in 7 cases, we followed thisguideline in the last patients who underwentsurgery.30,83-85
Fenestration
Hypertension in the cavopulmonary circuit or low systemic outputis frequently combined in the immediate post-operative period ofuniventricular surgery. Fenestration alleviates this hypertensionand increases cardiac output at the cost of decreasing systemicsaturation. This is a safety measure that allows time foradaptation to the new circulatory model, and in particularneutralizes the temporary deleterious effects of the surgicaltechniques on the cardiopulmonary system. This can be appliedselectively in the presence of risk factors, or as a survivalalternative when the previous MPAP is ≥19 mm Hg, or theintra-operative cavopulmonary gradient-FDVP is elevated. Our groupdid not habitually practice elective fenestration. In fact, it waselective in only 2 of 27 cases; the remainder were involuntary,with decreased early mortality in this subgroup (7.5%) being judgedas more important with respect to those who did not havefenestration (29.8%). Bridges et al showed that, when subgroupswith and without fenestration are compared, the latter have a lowerearly mortality rate, lower cavopulmonary pressures, and also lesstime with the thoracic drain in place. Other groups have alsoadvised fenestration of children with risk factors, arguing thatthe decrease in systemic oxygen saturation is compensated by theincrease in cardiac output which ultimately actually bringsr, dohedstudies that emfenestration. It appears that maintaless initialmortality in children with risk factors or with elevatedcavopulmonary pressures during surgery.11,86-88
Re-operation
Re-operations have been as frequent as 50% in survivors in olderseries; in the more recent series, re-operation frequency was notmore than 15%, and was greater in children who underwentatriopulmonary surgery vs cavopulmonary surgery. The incidence ofre-intervention in our series, in both surgical procedures andcatheterizations, is elevated, which is likely due to the 23-yearhistory of this type of surgery. Figure 9 points shows that only40% of the children were free of re-interventions. Thirty-fivepercent of the 95 survivors of the surgery required re-operation;the most frequent cause was hemodynamic (principally, residualshort circuit, subaortic stenosis, obstruction or stenosis of theatrial anastomosis, cavopulmonary stenosis, and tricuspid damage),and the second most common cause was arrhythmia (pacemakerimplant). In the subgroup of 33 re-operated patients, latemortality reached 36.3% (12 out of 33), while reaching only 13% (8of 62) in those who were not re-operated. Therefore, re-operationis a determining mortality factor in our series. We believe thatthis could be avoid by eliminating where possible the shunts orother palliative procedures prior to a bidirectional Glenn orFontan surgery, and perfecting a technique that prevents residuallesions or resorting to interventionalcatheterization.89-91
CONCLUSIONS
Surgery to achieve a univentricular circulatory model ispalliative and has a questionable quality of longevity. Thecavopulmonary technique involves less mortality, but long-termfollow-up is too short to differentiate the two. We must find newways to achieve the Fontan principle or explore other options thatimprove life expectancy and prognosis in children withuniventricular disease. Patient characteristics are the principledeterminants of mortality in this study, and therefore we mustprepare for this type of surgery in a rigorous manner, preservingthe functional reserve of the heart and lungs from the neonatalperiod on. Choussat was not wrong, and although the limits of whatcould be called the ideal univentricularization candidate have beenamplified, the same criteria still apply. The development of astrict risk index could contribute to more appropriate selectionand preparation of patients. Different strategies emerge as key inreducing immediate mortality; among these it is worth mentioningthe early use of partial cavopulmonary surgery, reduction orabolition of cardiopulmonary bypass and aortic cross-clamping,performing two-stage Fontan surgery, using fenestration in certaincases and developing a more precise and definitive technique toavoid re-interventions. It remains to be seen whether theseproposed safeguards will help reduce early mortality and at thesame time result in a more favorable prognosis at 3 or 4 decadesfollow-up, or on the contrary, result in the reconsideration ofother already know partial palliativetechniques.92-98
ACKNOWLEDGEMENT
The authors acknowledge the inestimable collaboration of Dr.Víctor Abraira, chief of the Biostatistic Clinic of ourhospital. We also thank our pediatric cardiology colleagues fromvarious locations promptly kept us informed of the course ofcertain patients.
NOTE
After this study was finished, we had the late death of 2 adultpatients: 1 who underwent the cavopulmonary technique who developeda septic infarct in the anterior descending coronary artery due tobacterial endocarditis, and the other who underwent atriopulmonarysurgery and died suddenly in the context of cardiac arrhythmiawhich required pacemaker implant, ventricular dysfunction, and AVvalve insufficiency.
Correspondencia:
Dr. M. Cazzaniga.
Servicio de Cardiología Pediátrica. HospitalRamón y Cajal.
Ctra. Colmenar, km 9,100. 28023 Madrid. España
E-mail: marcazza@inicia.es