TUESDAY MORNING, MAY 6, 1997

7:00 a.m. FORUM SESSION I - CARDIAC

Sheraton Ballroom

Moderators: Edward D. Verrier, M.D.

Pedro J. Del Nido, M.D.

F1. CREATION OF VIABLE PULMONARY ARTERY AUTOGRAFTS THROUGH TISSUE ENGINEERING.

John E. Mayer, Jr., M.D., Toshiharu Shinoka, M.D.*, Dominique Shum-Tim, M.D.*, Peter X. Ma, Ph.D.*, Ronn E. Tanel, M.D.*, Noritaka Isogai, M.D.*, Robert Langer, M.D.* and Joseph P. Vacanti, M.D.*

Boston and Cambridge, Massachusetts

Background. "Repair" of many congenital cardiac defects requires use of conduits to establish right ventricle to pulmonary artery continuity. Currently available homografts or prosthetic conduits lack growth potential and can become obstructed by tissue ingrowth or calcification leading to multiple conduit replacements. Tissue engineering (TE) is an approach where cells are grown in vitro onto biodegradable polymers to create "tissues" for implantation. A TE approach has recently been used to construct cardiac valve leaflets from autologous cells. This study assessed the feasibility of a TE approach to constructing pulmonary artery conduits.

Materials and Methods: Ovine artery (Grp A, N = 4) or vein (Grp V, N = 3) segments were harvested, separated into individual cells, expanded in tissue culture, and seeded onto synthetic biodegradable (polyglactin/polyglycolic acid) tubular scaffolds (20 mm long x 15 mm diameter). After 7 days in vitro culture the autologous cell/polymer vascular constructs were used to replace a 2 cm segment of main pulmonary artery in lambs (age = 68.4 ± 15.5d, weight = 18.7 ± 2.0 kg). One other animal received an acellular polymer tube sealed with fibrin glue. Animals were sacrificed at intervals of 11 to 23 weeks (mean follow-up = 125.4 30.8 days, mean weight 38.6 ± 13.0 kg) after echocardiographic and angiographic studies. Explanted TE conduits were assayed for collagen (4-hydroxyprolene) and calcium content, and a tissue DNA assay (bis-benzimide dye) was used to estimate number of cell nuclei. Mechanical tensile strength was evaluated with a vitrodyne V-1000 device.

Results: The acellular control (polymer only) graft developed progressive obstruction and thrombosis, but all 7 TE grafts were patent and demonstrated increase in diameter (Grp A = 18.3 ± 1.3 mm = 95.3% of native PA. Grp V = 17.1 ± 1.2 mm = 86.8% of native PA). None of the biodegradable polymer scaffold remained in any TE graft histologically. Collagen content in TE graft was 73.9 ± 8.0% of adjacent native PA. Tensile strength was 1.115 MPa (native PA = 0.583 MPa). Histologically elastin fiberswere present in the TE vessel wall and Factor VIII (specific for endothelium) was present on the luminal surface. DNA assay showed decreasing numbers of cell nuclei leftover 11 and 23 weeks suggesting an ongoing tissue remodeling. TE grafts calcium content was elevated (A= 7.95 ± 5.09, V= 13.2 ±5.48, native PA= 1.2 ±0.8 mg/g dry wt), but no macroscopic calcification was found.

Conclusion: In growing lambs vascular grafts engineered from autologous cells and biodegradable polymers functioned well in the pulmonary circulation and demonstrated increase in diameter and development of an extracellular matrix and an endothelial lining.

This tissue engineering approach may ultimately allow the development of viable vascular grafts for clinical use.

*By invitation

§F2. COMPARISON OF SURGICAL AND CATHETER-BASED TECHNIQUES OF VEGF DELIVERY ON MYOCARDIAL PERFUSION AND ENDOTHELIUM-DEPENDENT RELAXATION.

Frank W. Sellke, M.D., Motohisa Tofukuji, M.D., Ph.D.*, Roger Laham, M.D.*, Jianyi Li, M.B., M.S.*, Mukesh D. Hariawala, M.D.*, Stuart Bunting, M.D.*, and Michael Simons, M.D.*

Boston, Massachusetts and San Francisco, California

Previous studies have found that the administration of vascular endothelial growth factor (VEGF) in models of chronic myocardial ischemia significantly increases myocardial contractile function, in addition to increasing myocardial perfusion and coronary vascular endothelium-dependent relaxation, two major determinants in the development of unstable angina. In order to determine if surgically or catheter-based techniques of the administration of VEGF are superior at restoring myocardial perfusion and microvascular endothelium-dependent relaxation, ameroid occluders were placed around the left circumflex artery (LCx) of pigs. After 6 weeks, coronary angiography confirmed total LCx occlusion. VEGF was then administered to the epicardial surface of the LCx area with an implanted (thoracotomy) osmotic pump (20 mg over 3 weeks), via intracoronary (1C) injection (20 mg single bolus) through a LCx catheter, or via transvascular LCx injection (20 mg single injection). 1C injection of saline served as a control.

Myocardial blood flow (ml/min/gram tissue) in the collateral-dependent LCx territory and normally-perfused left anterior descending (LAD) artery territory was determined with colored microspheres. Arterioles (130 µm) were isolated from the LCx and LAD territories and examined in vitro with videomicroscopy. Arteriolar relaxations to the endothelium-dependent agonist adenosine 5' diphosphate (ADP) and the endothelium-independent vasodilator sodium nitroprusside (SNP) were studied in precontracted microvessels. Responses = % relaxation of U46619-induced contraction. *p<0.05 vs Control, †p<0.05 vs respective LCx value (2 way ANOVA and Fisher's test), n = 6 in each group. [Drug] = 10 uM in all cases.

While delivery of VEGF by a surgically implanted pump was associated with a return of myocardial perfusion to normal levels, it did not affect the impaired endothelium-dependent relaxation in the collateral-dependent LCx territory. Delivery of VEGF by either transvascular injection or intracoronary infusion was associated with improved myocardial blood flow, but also normalization of endothelium-dependent relaxation in the collateral-dependent territory. In conclusion, chronic myocardial ischemia is associated with decreased myocardial blood flow and reduced endothelium-dependent relaxation in the collateral-dependent coronary circulation compared to that in the normally perfused myocardium. Myocardial perfusion is restored with either surgically or catheter-based methods of growth factor delivery, while vascular reactivity is best restored with intravascular, catheter-based techniques.

§Authors have a relationship with Genentech

*By invitation

F3. NON-ANTICOAGULANT HEPARIN PRESERVES

REGIONAL MYOCARDIAL CONTRACTILITY AFTER ISCHEMIA-REPERFUSION INJURY: ROLE OF NITRIC OXIDE.

Peter C. Kouretas, M.D.*, Adam K. Myers, M.D.*, Young D. Kin, M.D.*, Jeff L. Myers, M.D., Ph.D.*, Yi-Ning Wang, M.D.*, Robert B. Wallace, M.D. and Robert L. Hannan, M.D.*,

Washington, DC

Prevention of myocardial dysfunction after ischemia-reperfusion (IR) injury remains a formidable challenge. We hypothesized that heparin may protect the myocardium from IR by a mechanism independent of its anticoagulant properties. Fifteen anesthetized dogs were subjected to 15 minutes ischemia followed by 120 minutes reperfusion and pre-treated with either saline (control, n = 5), heparin (6.0 mg/kg, n = 5) or N-acetylheparin (6.0 mg/kg, n = 5), a heparin derivative without anti-coagulant properties. The left anterior descending (LAD) artery was instrumented with an occluder and a pair of sonomicrometry crystals were placed in the myocardium for measurement of regional systolic shortening, a measure of myocardial contractility. Drugs or vehicle were administered after instrumentation and prior to LAD occlusion. The LAD was occluded for 15 minutes and functional recovery of myocardial performance was assessed at 15, 60 and 120 minutes reperfusion. In order to elucidate the role of the nitric oxide (NO) pathway, a specific NO inhibitor (nitro-L-arginine-1.5 mg/kg, n = 5) was given prior to heparin administration.

Systolic shortening was significantly depressed in the control group at 60 and 120 minutes reperfusion. Heparin and N-acetylheparin treated dogs, however, showed preservation of systolic shortening throughout reperfusion. Administration of the nitric oxide inhibitor nitro-L-arginine significantly attenuated heparin's protective effect on myocardial contractility during reperfusion. Activated clotting times were significantly elevated in the heparin and were normal in the N-acetylheparin and control groups. These results confirm the hypothesis that heparin preserves myocardial contractility after ischemia- reperfusion injury independent of its anticoagulant properties. Furthermore, the protective mechanism of heparin during ischemia-reperfusion injury appears to be regulated through the nitric oxide pathway. Administration of heparin derivatives may have important clinical implications in the prevention of myocardial injury without the adverse sequelae of bleeding.

*By invitation

F4. HYPOXIC INDUCTION OF TISSUE FACTOR PROMOTES ENHANCED ENDOTHELIAL CELL PRO-COAGULANT ACTIVITY.

Edward M. Boyle, Jr., M.D.*, Nigel Mackman, Ph.D.*, Timothy Pohlman, M.D.*, Timothy G. Canty, Jr., M.D.*, Owen Lawrence, Ph.D.* and Edward D. Verrier, M.D.

Seattle, Washington and La Jolla, California

The goal of treatment of myocardial ischemia is reperfusion, however, at times microcirculatory no-reflow contributes to ongoing ischemia once blood flow has been reestablished. Hypoxic endothelial cell activation may contribute to the no-reflow phenomenon, yet the mechanisms of impaired post-ischemic flow are poorly characterized. Because microthrombosis seems to play a role in this phenomenon we hypothesized that hypoxically activated endothelial cells would express the extrinsic pathway of coagulation activator, tissue factor (TF). TF is the most potent initiator of clotting known, and if expressed on the surface of ischemic endothelial cells it could result in impaired micro-circulatory blood flow upon reperfusion.

Methods: Cultured human umbilical vein endothelial cells (HUVEC) were exposed to normoxic (N) conditions (21% O₂) or hypoxia (H) in a controlled environmental chamber containing 2-3% oxygen for 2-24 hours. Additional HUVEC were exposed to H for 2 hours followed by 2-24 hours of reoxygenation in a normoxic environment (HR). N, H and HR cell lysates were assayed for TF promoter transcriptional activation by luciferase induction and TF protein production by Western blot analysis. The ability of these conditions to promote coagulation was assessed by exposing HUVECs treated with N, H, or HR to citrated human plasma in the presence of CaCU and recording the time to visible fibrin strand formation. TF activity per 10⁶ cells was plotted on a log-log curve against a standard curve constructed with various known concentrations of soluble tissue factor. The ability of a monospecific antibody to TF (aTF-ab) to inhibit this procoagulant activity is assessed to establish this response as secondary to TF.

Results: HUVECs treated with N alone do not make TF, as evidenced by a lack of constitutive promoter activity or TF protein on Western blots, however, following H or HR there is a 3-fold induction of TF promoter activity and a marked increase in TF protein manufactured by HUVECs. Functionally, there is a dramatic increase in procoagulant activity that peaks at 8 hours of H. (136 ± 54 vs. 8.4 ± 3.0) This response is markedly accenuated when cells were exposed to HR. (796 ± 511) Addition of aTF-ab completely abolishes the procoagulant responses to both H and H/R.

Conclusions: This work provides the first direct evidence that exposure of cultured HUVECs to H and HR increases the transcription, translation and surface expression of TF. Furthermore, addition of aTF-ab completely abolishes the potent procoaglulant response to H. Because of the extremely potent procoagulant response of human serum to TF it is conceivable that the expression of TF in vivo could contribute to impaired microcirculation after ischemia and reperfusion. The improved understanding of the role of the endothelial procoagulant response to ischemia/reperfusion should lead to more directed therapies to attenuate the post ischemic no-reflow phenomenon that contributes clinically to tissue injury and impaired myocardial function.

*By invitation

F5. ENDOTHELIAL DYSFUNCTION IN CEREBRAL MICROCIRCULATION DURING HYPOTHERMIC CARDIOPULMONARY BYPASS.

Pierantonio Russo, M.D.*, L. Craig Wagerle, Ph.D.* and Deborah A. Davis, M.D.*

Philadelphia, Pennsylvania

Sponsored by: Stanley K. Brockman, M.D., Philadelphia, Pennsylvania

Inflammatory stimuli and/or mechanical stresses associated with HCPB could potentially impair cerebrovascular function resulting in inadequate cerebral perfusion. We hypothesize that HCPB is associated with endothelial and/or vascular smooth muscle dysfunction and associated cerebral hypoperfusion. Therefore, we studied the cerebrovascular response to endothelium-dependent vasodilator, acetylcholine (Ach), endothelium-independent nitric oxide donor, sodium nitroprusside (SNP), and vasoactive amine, serotonin, in newborn lambs undergoing HCPB. Studies were performed on seven lambs equipped with a closed cranial window and cerebral arteriolar caliber (169 ± 22 µm diameter) was monitored using video microscopy. Topical application of Ach caused dose-dependent increase in diameter. This vasodilator response to Ach was absent in animals undergoing HCPB (left panel). HCPB did not alter the vasodilation in response to SNP (right panel). Furthermore, the contractile response to serotonin (10^-5 M) was fully expressed during HCPB (Adiameter = -29 ± 2 vs -30 ± 8%). The specific loss of Ach-induced vasodilation suggests endothelial cell dysfunction rather than impaired ability of vascular smooth muscle response to nitric oxide. It is speculated that loss of endothelium-ependent regulatory factors in the cerebral microcirculation during HCPB may enhance vasoconstriction and impaired cerebrovascular function may be a basis for associated neurological injury during or following HCPB.

*By invitation

F6. FLOW-INDUCED RELEASE OF EDRF DURING PULSATILE BYPASS: EXPERIMENTAL STUDY IN THE FETAL LAMB.

Gerard L. Champsaur, M.D., Catherine Vedrinne, M.D.*, Stephane Martinet, M.V.D.*, Franpois Tronc, M.D.*, Jacques Robin, M.D.* and Michel Franck, M.V.D.*

Lyon, France

Previous experimental studies have shown that when compared to continuous flow (CF) during fetal bypass, pulsatile flow (PF) enhances organs perfusion, particularly the placenta, through a diminution of vascular resistances. This study was initiated to test the hypothesis that fetal hemodynamic changes in this setting might be related to the release of endothelium-derived relaxing factor (EDRF) through oscillating shear stress and flow changes, as demonstrated in some isolated organ preparations.

Normothermic bypass was instituted in utero in 21 pre-term fetal lambs after maternal general anesthesia and usual hemodynamic instrumentation. In the fetus partially exposed through cesarean section, fetal bypass was established for a one-hour period after fetal sternotomy through right atrial and main pulmonary artery cannulations. Ultrasonic flowmeters were positioned around the post-ductal descending aorta and the umbilical artery. The circuit was primed with fresh blood and consisted of an oxygenator and a specific centrifugal pump set to either CF (n = 7) or PF (n = 7). Pump flow was monitored by an ultrasonic flowmeter placed around the pump outflow and was adjusted to maintain a physiological fetal main arterial pressure of 50 mmHg. EDRF blockade was carried out in seven animals (PBF) after 30 minutes of PF using a specific EDRF competitive inhibitor (N^ω-nitro--arginine) as a bolus followed by a continuous venous fetal infusion. Flows in ml/min expressed as mean ± SD were the following in each group after respectively 30 and 60 minutes of bypass.

Changes in systemic vascular resistances were similar, being significantly lower in Groups PF and PBF than in Group CF (550 ± 106 versus 821 ± 212 dynes/sec/cm" ). However, after EDRF blockade in Group PBF, resistances increased gradually to reach the level of that of group CF at the end of the bypass time (943 ± 77 versus 556 ± 143 dynes/sec/cm-5 in the non-blocked PF Group). In conclusion, EDRF blockade during 30 minutes returns fetal hemodynamics back to CF conditions. The specific EDRF inhbition agent used in this experiment suggests that nitric oxide may be released by fetal vascular endothelium during pulsatile bypass.

*By invitation

§F7. COMPLEMENT INHIBITION WITH SOLUBLE COMPLEMENT RECEPTOR TYPE I LIMITS ISCHEMIC DAMAGE DURING REVASCULARIZATION OF ACUTELY ISCHEMIC MYOCARDIUM.

Harold L. Lazar, M.D., Yusheng Bao, M.D.*, Samuel Rivers, B.S.*, TakafUmi Hamasaki, M.D.*, Sheilah Bernard, M.D.* and Richard J. Shemin, M.D.

Boston, Massachusetts

The increased inflammatory response resulting from complement activation during cardiopulmonary bypass (CPB) may contribute to myocardial damage during the revascularization of acutely ischemic myocardium. Soluble human complement receptor type 1 (SCR₁) is a recombinant form of human complement receptor which is a potent inhibitor of complement activation. This study was therefore undertaken to determine whether alteration of complement activation with SCR₁ would reduce myocardial dysfunction during the revascularization of acutely ischemic myocardium. In 20 pigs, the second and third diagonal coronary arteries were occluded for 90 minutes. Animals were then placed on CPB followed by 45 minutes of cold, antegrade, blood cardioplegic arrest and 180 minutes of reperfusion with the coronary snares released. In 10 pigs, SCR₁ (10 mg/kg) was intravenously infused over 30 minutes during the period of coronary occlusion; 10 other pigs received no SCR₁ (Unmodified). Total hemolytic complement activity (CH₅₀) was measured prior to ischemia, during coronary occlusion and reperfusion and expressed as the percent of preischemic values. Ischemic damage in the area at risk was assessed by measuring the change in myocardial tissue pH (ApH) from preischemic values; Wall Motion Scores (WMS) using transthoracic echo-cardiography (4 = normal to -1 = dyskinesia) and infarct size (Area of Necrosis/Area at Risk; AN/AR) using histochemical staining. Data is expressed as the Mean ± Standard Error.

We conclude that complement inhibition with SCR₁ significantly limits ischemic damage during the revascularization of acutely ischemic myocardium.

§ Presenter has a relationship with T Cell Sciences, Inc.

*By invitation

F8. EXTRACELLULAR SUPEROXIDE DISMUTASE TRANSGENE OVEREXPRESSION SIGNIFICANTLY IMPROVES PRESERVATION OF MYOCARDIAL FUNCTION FOLLOWING ISCHEMIA AND REPERFUSION INJURY.

Edward P. Chen, M.D.*, Hartmuth B. Bittner, M.D., Ph.D.*, R. Duane Davis, M.D.*, Peter Van Trigt, M.D. and Rodney J. Folz, M.D., Ph.D.*

Durham and Greensboro, North Carolina

Myocardial injury after ischemia and reperfusion injury may be mediated, at least in part, by oxygen-derived free radicals and is supported by the observation that significant quantities of these radicals are generated during post-ischemic reperfusion. To directly assess the protective effect of the extracellular superoxide dismutase (EC-SOD), a controlled prospective, double-blinded experimental study was performed to evaluate myocardial function in the hearts of transgenic mice overexpressing human EC-SOD to levels 3.5x greater than controls. Heterozygous (EC-SOD, n = 6, 22-26 g) and nonheterozygous litter mate controls (CTL, n = 8, 22-26 g) were analyzed by PCR analysis of tail DNA. An isolated work-performing murine heart preparation was used to evaluate preload-dependent cardiac output (CO), contractility (dP/dt), stroke work (SW), stroke volume (SV), and heart rate (HR) before (Pre-I) and after (Post-I) a 6 minute period of normothermic ischemia. Results are expressed as mean ± SEM (ANOVA, paired/unpaired t-test). There was no significant difference between EC-SOD and CTL in any parameter of myocardial function Pre-I. The average Pre-I HR for CTL and EC-SOD was 438 ± 19 beats/min and 482 ± 16 beats/min. There was an 87% recovery in post-I HR in CTL and a 94% recovery in post-I HR in EC-SOD (p<0.05). Pre-I SW/SV/ dP/dt in CTL were 674 + 79 dyne*cm/12.7± 1.6 µl/2305± 157 mmHg/s, while Pre-I EC-SOD SW/SV/ dP/dt were 593 ± 28 dyne*cm/10.6 ± 0.6 µl/ 2127 ± 104 mmHg/s. Post-I SW/SV/ dP/dt in CTL recovered by 55%/54%/80%, while Post-I EC-SOD SW/SV/ dP/dt recovered 77%/78%/90% (p<0.001). The table displays the preload dependent Frank-Starling relationships in CO (mL/min) in both groups Pre-I and Post-I (*=p<0.01 Pre-I vs Post-I; t = P<0.05 EC-SOD vs CTL, SEM in parentheses):

Conclusions: EC-SOD transgene overexpression does not affect baseline myo-cardial function compared to CTL hearts. Following global normothermic ischemia and subsequent reperfusion, significant decreases in cardiac function were observed in both EC-SOD and CTL, however, a significantly higher percentage of recovery was observed in EC-SOD overexpressed hearts. These data suggest that EC-SOD transgene overexpression significantly improves preservation of myocardial function following ischemia and reperfusion injury.

*By invitation

9:00 a.m. PLENARY SCIENTIFIC SESSION

Sheraton Ballroom

Moderators: David B. Skinner, M.D.

James L. Cox, M.D.

14. OPERATIVE OUTCOME AND HOSPITAL COST.

Victor A. Ferraris, M.D., Ph.D., Suellen P. Ferraris, Ph.D.* and Amandeep Singh*

Albany, New York

Discussant: Floyd D. Loop, M.D.

Introduction: Health care costs are increasing at an alarming rate and cardiac procedures contribute to this increase. It is likely that patient risk factors contribute to this cost increase since operative interventions are being performed on high risk patients with greater frequency, but the exact relationship of patient risk factors to hospital cost is poorly understood. Because of this knowledge deficit and because of the possibility that modification of patient risk factors might lead to decreased cost, we undertook a study to identify patient risk factors associated with increased hospital cost and to evaluate the relationship of increased cost to serious hospital morbidity and mortality.

Methods: More than 100 patient variables were collected in a prospective manner in 1221 patients undergoing cardiac procedures. Simultaneously, patient hospital cost was computed from the cost-to-charge ratio after validation of the individual departmental ratios. Univariate statistics were used to explore the relationship between hospital cost and patient outcomes. Multivariate regression models using logistic regression, Cox proportional hazards regression, and stepwise linear regression identified independent patient risks for mortality, hospital morbidity (inferred from length-of-stay) and hospital cost, respectively.

Results: The greatest cost occurred in 31 patients who did not survive operation ($74, 466 ± 19, 393 95% CI). This was significantly greater than the cost in 120 patients who suffered serious non-fatal morbidity ($60, 335 ± 6, 248 95% CI, p = 0.02) and the cost in 1070 patients who survived operation without complication ($31, 459 ± 711 95% CI, p<0.01). Hospital cost was not directly related to length of stay (LOS), although the increased cost in operative fatalities was associated with increased length of stay compared to uncomplicated procedures (14.2 ± 7.4 95% CI vs. 8.3 ± 0.3 95% CI days, p = 0.02). Breakdown of the components of hospital cost in fatalities and in patients with non-fatal complications revealed that the greatest contribution to cost was in the anesthesia and operating room costs as well as pharmacy costs, two components not directly related to LOS. Significant independent risks for mortality, morbidity (LOS), and hospital cost identified by multivariate regression are shown in the table:

Conclusions: We conclude that: 1) operative death is the most costly outcome, 2) LOS is not an accurate indicator of hospital cost, 3) ventricular dysfunction associated with operations for other than coronary disease are significantly associated with increased cost, and 4) patient factors that are amenable to preoperative intervention to reduce costs are correction of preoperative anemia (i.e. increase RBC volume) and treatment of CHF. These results suggest a high-risk patient profile that should be a target for cost reduction strategies.

*By invitation

15. EN BLOC ESOPHAGECTOMY IMPROVES SURVIVAL FOR STAGE III ESOPHAGEAL CANCER.

Nasser K. Altorki, M.D., Leonard Girardi, M.D.* and David B. Skinner, M.D.

New York, New York and Houston, Texas

Discussant: Victor F. Trastek, M.D.

The role of enbloc esophagectomy in patients with locally advanced esophageal cancer is not well defined. Between January 1988 and June 1992 we continued our selective surgical approach whereby patients with favorable disease (Stage I and II) were treated by enbloc resection while patients with suspected Stage III disease underwent resection by standard techniques. Since the mortality and morbidity of both techniques appeared similar we adopted enbloc esophagectomy more liberally in all stages since 1992. The purpose of this study was to examine the influence of this strategy on the survival of resected Stage III patients.

Between January 1988 and September 1996, 128 patients underwent esophagectomy by an enbloc technique (n = 78) or a standard esophagectomy (n = 50, 46 transthoracic, 4 transhiatal). There were 101 males and 27 females with a median age of 62 (range 34-87). Squamous cell cancer was present in 38 and adenocarcinoma in 90 patients. Hospital mortality was 5.4% (7/128) and morbidity 54% and was not influenced by the type of procedure. Enbloc resection was done in 100% of Stage 0 and Stage I patients (18), 66% of Stage II patients and 60% of Stage III patients. Two and three year survival, median survival and p values are shown below for all stages excluding Stage IV (n = 23).

Eighty-six patients had positive nodes (68%). Two and three year survival for patients with Nl disease treated by enbloc resection was 59% and 35% respectively (median 23 m) versus 13% and 6% (median 12.6 m) for patients treated by standard resections, (p = 0.007).

Based on a significant improvement in 2 year and 3 year survival and median survival we conclude that enbloc resection improves survival in patients with Stage III cancer of the esophagus and that excision of nodal disease may have a positive survival benefit.

*By invitation

16. GROWTH POTENTIAL AFTER BIVENTRICULAR REPAIR IN CHILDREN WITH SMALL BUT NOT HYPOPLASTIC LEFT HEART.

†Alain E. Serraf, M.D., Nicolas Bonnet, M.D.*, François Lacour-Gayet, M.D.*, Dominique Piot, M.D.*, Anita Touchot, M.D.*, Jacqueline Bruniaux, M.D.* and Claude Planché, M.D.

LePlessis-Robinson, France

Discussant: Thomas L. Spray, M.D.

Because of the lack of strict criterias, there is still no agreement whether to perform uni or biventricular repair in children with small left heart structures. Thirty six children with small left heart sizes underwent biventricular repair at our Institution. Preoperative echocardiographic assessment allowed to record the diameter of the mitral valve, end-diastolic (EDLVD) and end-systolic (ESLVD) left ventricular diameters, aortic and subaortic root diameters. Left ventricular volumes (EDLVV and ESLVV) were estimated according to the corrected formula and stroke volumes (SV) were then calculated. All measurements were standardized to normal by the Z-value method. All the pts presented with small but normal left ventricular anatomy, those with profound structural anomalies of the left ventricle (LV) (hypoplastic left heart syndrome, complete AV canals) were excluded from this work. There were 19 males and 17 females. The median age at the first operation was: 13 days and the mean weight was 3.5 ± 0.9 kg. Three groups could be distinguished: Group I (n = 15) with high left to right atrial shunt and insufficient preload of the LV, Group II (n = 9) with elevated left ventricular afterload and Group III (n = 12) which combined both anomalies of pre and afterload. Sixteen underwent single stage complete repair and 20 had an incomplete or palliative procedure. Early reoperation was necessary in 12 pts because of the inability of the LV to sustain systemic output. Two were converted to univentricular first stage palliation and the other underwent closure of atrial shunts to preload the LV. There were 6 early deaths, 5 of whom could be attributed to the inability of the LV to sustain a systemic output. All the survivors demonstrated a rapid growth of left heart structures already at discharge from hospital.

Statistical analysis demonstrated that non survivors had a smaller aortic root diameter (p=0.01) Eleven patients underwent 20 late reoperations for closure of residual ventricular shunts (n = 6), subaortic stenosis (n = 3) and for mitral valve stenosis (n = 2) with 2 deaths. At the time of reoperations, the sizes of left heart structures were within normal ranges in all but 2 patients. A median follow-up of 40 months (Ranges: 6-120 months) was achieved in all survivors. They were all in NYHA classes HI with normal sizes of left heart structures. Actuarial survival and freedom from reoperation rates at 8 years were 60.6 ± 15.9% and 25.2 ± 11.2%. In conclusion, biventricular repair promotes rapid growth of the left ventricular structures in children with small but not hypoplastic left heart and appears to be a good surgical option. It remains however difficult to determine clear predictive criterias for either biventricular or univentricular repair.

10:00 a.m. INTERMISSION - VISIT EXHIBITS

†1993-94 AATS Graham Fellow

*By invitation

10:45 a.m. PLENARY SCIENTIFIC SESSION

Sheraton Ballroom

Moderators: David B. Skinner, M.D.

James L. Cox, M.D.

17. INDUCTION THERAPY FOR ESOPHAGEAL CANCER WITH PACLITAXEL (TAXOL®) AND HYPER-FRACTIONATED RADIOTHERAPY: A PHASE I/II STUDY.

Cameron D. Wright, M.D.*, John C. Wain, M.D.*, Thomas J. Lynch, M.D.*, Noah C. Choi, M.D.*, Hermes C. Grille, M.D. and Douglas J. Mathisen, M.D.

Boston, Massachusetts

Discussant: Michael E. Burt, M.D.

Induction chemoradiation followed by surgery is a promising approach to treatment for esophageal cancer. Previous reports emphasize the importance of a high pathologic complete response rate as these patients have enhanced survival. Paclitaxel (Taxol ) is a new agent with high response rates in metastatic esophageal cancer. However, the use of Paclitaxel has not been reported in induction regimens. Twenty-seven patients with esophageal cancer were enrolled in a Phase I/II trial of induction chemoradiation followed by esophagectomy beginning in May 1995. The chemotherapy consisted of paclitaxel at 3 dose levels (75, 125 and 100 mg/m²), cisplatin and 5-fluorouracil. The radiotherapy was concurrent and hyperfractionated and delivered 42 Gray to the mediastinum with a 16.5 Gray boost to the tumor. The mean age of the patients was 60 and 21 (78%) had adenocarcinoma. Pretreatment staging was by computed tomography and endoscopic ultrasonography. Nine patients were T2NO, 1 T2N1, 11 T3NO, 5 T3N1, and 1 was T4NO. Patients were hospitalized for a mean of 15.8 days for chemotherapy or complications of induction treatment. The number of patients who had severe (Grade 4) esophagitis at each dose level of paclitaxel was as follows: 75 mg/m²-50%, 125 mg/m²-75%, and 100 mg/m²-47%. One patient died during induction therapy at home presumably from sepsis. Twenty-six patients underwent esophagectomy with a mean hospital stay of 13.7 days. The average time to surgery from the initiation of treatment was 77 days. There was 1 postoperative death due to an aorto-esophageal fistula. Eleven of 26 patients (42%) had a complete pathologic response in the resected specimen. Nineteen of 26 patients (73%) had no tumor in the resected nodes. Twelve of 26 patients (46%) had no tumor in the resected esophagus. Two patients have recurred with distant disease and died. Twenty-three patients have no evidence of disease with a mean follow-up of 10.3 months. In this regimen, paclitaxel at a dose of 100 mg/m² appears to have acceptable toxicity. Previous reports suggest a pathologic complete response rate up to 25% with most induction regimens. The relatively high pathologic complete response rate (42%) with this regimen is encouraging but survival data are not yet available to confirm increased survival.

*By invitation

18. EARLY RESULTS WITH PARTIAL LEFT VENTRICULECTOMY.

Patrick M. McCarthy, M.D., Randall C. Starling, M.D.*, Gregory M. Scalia, M.B.B.S.*, James D. Thomas, M.D.*, Nicholas G. Smedira, M.D.* and James B. Young, M.D.*

Cleveland, Ohio

Discussant: D. Craig Miller, M.D.

Partial left ventriculectomy (PLV), the Batista procedure, has demonstrated significant clinical improvement in some patients (pts) with dilated cardiomyopathy (DCM). Since May 1996 we have performed PLV in 30 patients, initially in heart transplant (Tx) candidates, and more recently in non-Tx candidates. The mean age of the pts was 54 years (range 34 to 72); 60% were Class IV and 40% Class III. Preoperatively all pts were thought to have idiopathic DCM. As our experience has accrued we have increased the extent of left ventriculectomy and more complex mitral valve (MV) repairs. For only one pt was MV replacement performed (rheumatic MV disease). For 29 pts the anterior and posterior MV leaflets were approximated (Alfieri repair); 24 pts also had ring posterior valvuloplasty. The lateral wall (circumflex territory) between the papillary muscles was the location for ventriculectomy in 29 pts. In 4 pts the posterior papillary muscle was divided, additional posterior wall was resected, and the papillary muscle heads reimplanted. The extent of resection was gauged by a formula using the LV internal diameter (LVID) and interpapillary muscle distance (IPD) to predict post resection diameter:

LVID_post=LVID_pre- (IPD / p)

All ventriculotomies were closed with soft felt or bovine pericardium. Intraoperative hemodynamic and echocardiographic changes for 17 pts operated before October are in the Table. Initial pressure area loops have shown decreased stroke work, filling pressures, and filling volumes; with preserved stroke volume.

There were no in-hospital deaths; there was one reoperation for bleeding. Three pts required HeartMate left ventricular assist devices postoperatively; two are being weaned as the heart recovers. Three pts have been relisted for Tx. There was one death at 3 months from cerebral edema. Of 20 discharged pts 80% are subjectively Class I or II.

Conclusion Ninety-seven percent of pts with severe LV dysfunction and mitral regurgitation are alive following PLV and valve repair; most are clinically improved. The clinical outcome is not always predictable however, and therefore longer follow-up and further studies are required to optimize pt selection and surgical techniques.

11:25 a.m. ADDRESS BY HONORED SPEAKER

Esophageal Surgery at the End of the Millenium.

Antoon E.M.R. Lerut, M.D., Leuven, Belgium

12:10 p.m. ADJOURN FOR LUNCH - VISIT EXHIBITS

12:10 p.m. CARDIOTHORACIC RESIDENTS' LUNCHEON

*By invitation