Mixed Venous Oxygen Saturation Monitoring After Stage 1 Palliation for Hypoplastic Left Heart Syndrome

Transcription

1 Mixed Venous Oxygen Saturation Monitoring After Stage 1 Palliation for Hypoplastic Left Heart Syndrome James S. Tweddell, MD, Nancy S. Ghanayem, MD, Kathleen A. Mussatto, BSN, Michael E. Mitchell, MD, Luke J. Lamers, MD, Ndidiamaka L. Musa, MD, Stuart Berger, MD, S. Bert Litwin, MD, and George M. Hoffman, MD Herma Heart Center and Children s Research Institute, Children s Hospital of Wisconsin, Division of Cardiothoracic Surgery, Department of Surgery, The Section of Critical Care and Cardiology, Department of Pediatrics, and the Department of Anesthesia, Medical College of Wisconsin, Milwaukee, Wisconsin Background. Staged palliation for hypoplastic left heart syndrome has been marked by high early mortality due to the limited cardiac output of the postischemic single right ventricle combined with the inefficiency and volatility of parallel circulation. Methods. Since July 1996, we have performed stage 1 palliation (S1P) in 178 patients. Within this group is a consecutive cohort of 116 patients with true hypoplastic left heart syndrome that underwent S1P with a modified Blalock-Taussig shunt. A prospective database containing postoperative hemodynamic data was maintained on all patients. Studied were the incidence of organ failure, extracorporeal membrane oxygenation (ECMO), and mortality, as well as the relationship between these outcomes and postoperative hemodynamics. Results. Hospital survival for this cohort was 93% (108/116). Patients who died after S1P had a lower superior vena cava oxygen saturation (SvO 2 ) level compared with survivors (53.1% 10.6% versus 59.3% 9.2%, p 0.034). Renal failure developed in 2 (1.7%) of the 116 patients, necrotizing enterocolitis developed in 1 (0.9%), and 5 (4.3%) had clinical seizures. ECMO support was instituted in 12 patients (10.3%). The SvO 2 level was lower in patients requiring ECMO (54.0% 9.7% versus 59.9% 9.2%, p 0.031). Conclusions. Goal-directed therapy with SvO 2 as an indicator of systemic oxygen delivery is associated with excellent early survival and a low incidence of organ failure after S1P. Inability to optimize SvO 2 in the early postoperative period is associated with an increased risk of organ failure, ECMO, and death. (Ann Thorac Surg 2007;84: ) 2007 by The Society of Thoracic Surgeons The early postoperative period after the Norwood procedure is one of high risk due to inherent inefficiencies of parallel circulation and the inferior power source of the postischemic single right ventricle. With limited cardiac output and parallel circulation, alteration of the pulmonary to systemic flow ratio (Qp/Qs) in either direction can result in critical reduction of systemic oxygen delivery. Historically, monitoring during this period has included physical exam, blood pressure, heart rate, central venous pressure (CVP), and measurement of systemic arterial oxygen saturation (Sao 2 ) using pulse oximetry. We hypothesized that objective measurement of systemic oxygen delivery would permit better management of the patient during this vulnerable period after the Norwood procedure. For the last 10 years we have used continuous venous oximetry from the superior vena cava Accepted for publication May 1, Presented at the Forty-third Annual Meeting of The Society of Thoracic Surgeons, San Diego, CA, Jan 29 31, Address correspondence to Dr Tweddell, Children s Hospital of Wisconsin, 9000 W. Wisconsin Ave, Milwaukee, WI 53226; jtweddell@chw.org. (Svo 2 ) as an objective assessment of systemic oxygen delivery and as a means to guide management. This study reviews our experience using Svo 2 monitoring to manage this complex group of patients. Patients and Methods Patients As of June 2006, 178 consecutive patients represent a single-center, 10-year experience with staged palliation for all patients with hypoplastic left heart syndrome (HLHS) or other variants of congenital heart disease with systemic outflow obstruction. This time period represents the use of continuous Svo 2 monitoring for the postoperative management of patients who underwent stage 1 palliation (S1P). As previously described, a 4F oximetric catheter (Abbot Laboratories, North Chicago, IL) was placed routinely through a small incision directly into the superior vena cava positioned 2 to 3 mm within the vessel [1]. The focus of this review is 116 of the 178 patients with HLHS as defined by International Working Group for Mapping and Coding of Nomenclatures for Paediatric 2007 by The Society of Thoracic Surgeons /07/$32.00 Published by Elsevier Inc doi: /j.athoracsur

2 1302 TWEDDELL ET AL Ann Thorac Surg OXIMETRY MONITORING AFTER STAGE 1 PALLIATION 2007;84: Table 1. Complications Organ System Complication Patients, n (%) Total Patients per Organ System, n (%) Circulatory Arrhythmia 16 (13.8) 45 (38.8) Congestive heart failure 19 (16.3) ECMO 12 (10.3) CPR 12 (10.3) Heart transplant 1 (0.9) Renal Dialysis 2 (1.7) 5 (4.3) Oliguria, hyperkalemia 3 (2.5) Gastrointestinal Necrotizing enterocolitis 1 (0.9) 1 (0.9) Neurologic Clinical seizures 5 (4.3) 5 (4.3) Intraventricular hemorrhage 0 Stroke 1 CPR cardiopulmonary resuscitation; ECMO extracorporeal membrane oxygenation. and Congenital Heart Disease, the Nomenclature Working Group, who underwent S1P with a modified Blalock- Taussig shunt [2]. The study excluded 37 patients who underwent S1P for hypoplastic left heart variants, specifically single-ventricle anatomy with systemic outflow obstruction with arch obstruction such as single left ventricle with transposed great vessels, unbalanced atrioventricular canal, and double-outlet right ventricle with mitral atresia. Also excluded were 25 patients who met the definition of HLHS but underwent S1P with a right ventricle to pulmonary artery conduit. The aim was to isolate a cohort with a uniform diagnosis undergoing a standardized operation. Operative Technique and Postoperative Management The operative technique and postoperative management have been previously described in detail [1, 3 5]. Highdose aprotinin was used in all patients and 110 received phenoxybenzamine. Cardiopulmonary bypass (CPB) strategies for arch reconstruction included predominately deep hypothermic circulatory arrest (DHCA) in 54 patients and continuous cerebral perfusion in 62. In general, arch reconstruction included a coarctectomy and augmentation of the arch and ascending aorta using pulmonary homograft. Shunt sizing was based on the patient s weight and age. Patients who weighed less than 3.2 kg generally received a 3.0-mm shunt, and those heavier than 3.2 kg received a 3.5-mm shunt. For patients who presented after 2 weeks of life, a larger shunt was used. Modified ultrafiltration and delayed sternal closure were routine. Monitoring and Data Collection Standardized postoperative monitoring for this cohort included heart rate, continuous invasive arterial blood pressure, central venous pressure (CVP), Sao 2,Svo 2, and end-tidal carbon dioxide. Postoperative management targets of Sao 2 exceeding 75%, Svo 2 exceeding 50%, mean arterial blood pressure (MAP) exceeding 45 mm Hg, diastolic blood pressure exceeding 30 mm Hg, and hematocrit exceeding 45% were achieved by titration of vasoactive drugs, red cell transfusion, analgesia, sedation, and controlled ventilation. This study was approved by the Children s Hospital of Wisconsin Human Research Review Board and was conducted in accordance with all human research regulatory requirements. Informed written consent was obtained from the parents of the most recent 110 patients reported in this study for prospective data collection of hemodynamic and outcome data after surgery for HLHS. The first 6 patients in this study had prospective collection of hemodynamic data in 1996 before human studies approval was generally required for reporting of anonymous data. The Children s Hospital of Wisconsin Human Research Review Board approved the inclusion of these 6 additional patients and waived the need for parental consent. For all patients, recording of Svo 2,Sao 2, MAP, heart rate, CVP, and inspired oxygen fraction (Fio 2 ) are made upon arrival in the intensive care unit and hourly for 48 hours. Additional data collected during the first 48 postoperative hours include hemoglobin level, arterial partial pressure of carbon dioxide (Paco 2 ), arterial partial pressure of oxygen (Pao 2 ), and inotropic support. Derived variables include pulmonary-to-systemic flow ratio (Qp/ Qs), arterial-venous oxygen content difference ( AVo 2 ), the indices for systemic (SVRI) and pulmonary vascular resistance (PVRI), assuming an oxygen consumption of 160 ml/(m 2 min), and ventilation index [ventilator rate (peak inspiratory pressure end expiratory pressure) Paco 2 /1000]. Demographics, preoperative support, intraoperative support, shunt size, and outcome data were available for all patients. The impact of a variety of factors (Appendix) on outcome end points was determined. International Classification of Diseases, 9th revision (ICD-9), diagnostic and procedural coding for all patients at time of discharge after S1P was reviewed for death and evidence of postoperative complications. Complications were categorized by organ system and are summarized in Table 1. Validity of diagnosis and procedure coding was confirmed by a review of all medical records. One of three end points

3 Ann Thorac Surg TWEDDELL ET AL 2007;84: OXIMETRY MONITORING AFTER STAGE 1 PALLIATION 1303 was assigned to each patient: uncomplicated survival, survival with complications (as defined in Table 1), or early death (mortality within 30 days of S1P or within the primary S1P hospitalization). Statistical Analysis Statistical analysis was completed using SPSS Advanced Models 13.0 software (SPSS, Inc, Chicago, IL) and Stata software (StataCorp, College Station, TX). Analysis of impact of identified factors on outcome was completed for continuous variables using analysis of variance (ANOVA) and time series regression, and 2 and the likelihood ratio test were used for categoric variables. Adjustment for multiple comparisons was performed using the Tukey honestly significant difference method with a significance cutoff at p Multivariable analysis was performed to determine factors related to outcome using stepwise ordered logistic regression with a threshold for rejection at p 0.2 and cutoff for significance at p Actuarial survival analysis was performed using Kaplan-Meier methods. Data are reported as mean standard deviation for continuous variables and count with percentage for categoric variables. Time series regression techniques were used to derive timeweighted mean values. Median values are included where appropriate. Results Aortic atresia was present in 68% of patients (79/116), with associated mitral valve atresia in 58 and a patent mitral valve in 21. Aortic stenosis occurred in 32% (37/ 116), and a restrictive ventricular septal defect was present in 11% of the patients (13/116) with aortic stenosis. The native ascending aorta measured mm, and 41% (48/116) had a native aortas of less than 2.5 mm. Boys comprised 65% of the patients (75/116), and 43% (75/116) were diagnosed in utero, with prenatal diagnosis occurring more frequently in recent years. Mean age at S1P was days (median 5 days, range, 2 to 60 days), and mean weight was kg (range, 1.7 to 4.6 kg). Nine percent (11/116) were less than 37 weeks gestational age (mean, weeks; range, 31 to 36 weeks), and 5% (6/116) weighed less than 2.5 kg at birth (range, 1.7 to 2.4 kg). Extracardiac anomalies were identified in 12% (14/116), and additional cardiac diagnoses were present in 7% (8/116). Specifics of the additional diagnoses are summarized in Table 2. Total intraoperative support time (CPB plus DHCA) averaged minutes (range, 99 to 358 minutes). The strategy in 58% (67/116) was continuous cerebral perfusion (CCP) during arch reconstruction, with DHCA times that were significantly lower for these patients than for those undergoing arch reconstruction using only DHCA (13 14 minutes versus minutes, p 0.001). The total support times were nearly identical between the DHCA group and the CCP group ( minutes versus minutes, p 0.991). Delayed sternal closure was used in all but 2 patients. The median time to delayed sternal closure was 3 days (mean, 3.9 Table 2. Additional Diagnoses Diagnosis Patient, n Cardiac Infant of a diabetic mother - Hypertrophic 1 cardiomyopathy Aortic root thrombus 2 Transmural infarct (preoperative) 1 Anomalous origin of a coronary artery 2 from the pulmonary artery Coronary sinusoids 1 Intact atrial septum a 1 Partial anomalous pulmonary venous 1 drainage Extracardiac Tracheoesophageal fistula with 1 diaphragmatic hernia Vertebral anomalies 1 Partial duplication chromosome 15 1 Suspected mitochondrial myopathy 1 Turner syndrome 1 Cystic fibrosis ( F508 homozygous) 1 Structural renal anomalies 3 Unknown syndrome (cataracts, 1 dysmorphic facies) Ambiguous genitalia or urologic tract 2 abnormalities Immune deficiency 2 a Patient with intact atrial septum also had aortic root thrombus. 2.5 days; range, 0 to 14 days). The median hospital length of stay was 29 days (mean, days; range, 10 to 135 days). Figure 1 outlines the surgical progress and outcomes for these patients. There were eight (6.9%) early deaths. Mean time to death after S1P was days (range, 10 to 50 days), and the mean age at death was days (range, 15 to 68 days). Seven of the 8 patients who died had significant postoperative complications, and 3 had multiple complications. Causes of early death included cardiac failure in 6, shunt obstruction in 1, and sudden death at home while feeding in 1 patient on day 29 of life. No patient with low birth weight, prematurity, or extracardiac anomalies died early after S1P. Of the S1P survivors, 10 patients died before stage 2 palliation (S2P), 1 received a cardiac transplant, and 97 progressed to S2P with bidirectional cavopulmonary anastomosis. After S2P with bidirectional cavopulmonary anastomosis, there was one early death and three late deaths. Two patients received transplants and are survivors. To date, 75 patients have progressed to completion Fontan with no takedowns, and 16 patients are awaiting completion Fontan. No patient died early after the Fontan operation; however, 1 patient died late and another received a cardiac transplant. Follow-up was complete on 100% of patients to a mean age of months (median, 59.8 months; range,

4 1304 TWEDDELL ET AL Ann Thorac Surg OXIMETRY MONITORING AFTER STAGE 1 PALLIATION 2007;84: Fig 1. The outcome of 116 patients with hypoplastic left heart syndrome (HLHS) undergoing a Norwood procedure using a systemic to pulmonary artery shunt. (BDG bidirectional Glenn; BT Blalock-Taussig.) 0.5 to months). Actuarial survival for this cohort of 116 patients is 80% at 5 years and is shown in Figure 2. An end-organ complication occurred in 47 (41%) of 116 patients after S1P, of whom 40 went on to be early survivors. Several patients had multiple complications. Extracorporeal membrane oxygenation (ECMO) support was required in 12 patients (10.3%) during S1P hospitalization, of which 2 underwent ECMO cannulation in the operating room for failure to wean from CPB, and both survived S1P. Between 1997 and 2000, 5 patients were cannulated for cardiac failure independent of shunt occlusion, 3 of whom required cardiopulmonary resuscitation (CPR) just before cannulation. Two patients with evidence of low cardiac output syndrome were cannulated without CPR. Three of the 5 patients were decannulated but died of progressive cardiogenic shock 7, 14, and 16 days later. Two of the 5 patients had support withdrawn while on ECMO owing to persistent multiorgan failure. Fig 2. Kaplan-Meier survival for all 116 patients undergoing stage 1 palliation of hypoplastic left heart syndrome with a systemic to pulmonary artery shunt. Between 1998 and 2005, 5 patients were supported with ECMO due to acute shunt occlusion, of which 3 received CPR before cannulation, and 4 survived S1P. Overall S1P mortality for patients placed on ECMO was 50%, and all ECMO survivors have undergone successful completion Fontan. CPR was required in 12 patients (10.3%) during S1P hospitalization: 4 as a result of acute shunt occlusion, 1 after a respiratory arrest, and 7 for cardiac failure. Five of the 7 patients who received CPR for cardiac failure are included in the group who went on to ECMO before Two additional patients had CPR and were not placed on ECMO: 1 as a result of a hyperkalemic arrest and 1 for cardiac tamponade. Six (50%) of the 12 patients who received CPR survived S1P; 4 had completion Fontan and 2 underwent cardiac transplantation. Most of the postoperative complications were within the circulatory system. Additional complications included renal failure requiring dialysis in 2 patients (1.7%), renal insufficiency necessitating treatment for oliguria with hyperkalemia in 2 (2.5%), necrotizing enterocolitis in 1 (0.9%), and clinical seizures in 5 (4.3%). One patient with seizures had a documented embolic stroke. These results are summarized in Table 1. The results of univariate analysis are summarized in Table 3. Complications were associated with longer duration of intraoperative cardiopulmonary support, wider AVo 2, lower MAP, lower ph, and lower base deficit. A borderline association was found between complications and higher Qp/Qs. Mortality was associated with a lower Svo 2, higher hemoglobin level, and wider AVo 2. There was a borderline association between the use of a higher Fio 2 and mortality. The results of the multivariable analysis are summarized in Table 4. Complications were predicted by younger age, longer duration of intraoperative cardiopulmonary support, lower Svo 2, higher Sao 2, lower heart rate, higher Fio 2, and lower ph. Only lower Svo 2 predicted death. Poor outcome incorporating both complications and death was predicted in an ordered logistic

5 Ann Thorac Surg TWEDDELL ET AL 2007;84: OXIMETRY MONITORING AFTER STAGE 1 PALLIATION 1305 Table 3. Results of Univariable Analysis of Impact of Factors on Outcome Survived (n 68) Factor All (n 116) Uncomplicated (n 68) Complications (n 40) Died (n 8) p Value Age at operation (days) Weight (kg) Gestational age (weeks) Prenatal diagnosis (n) Aortic diam (mm) Aortic diam 2.5 mm (n) Male gender (n) Aortic atresia (n) Additional diagnoses (n) Pre-op ventilation (n) Phenoxybenzamine (n) CCP (n) Total support time a (min) Shunt size normalized (mm 2 /m 2 ) Svo 2 (%) b a Sao 2 (%) Hemoglobin (mg/dl) b a 0.03 AVo 2 (ml/dl) b a, b a MAP (mmhg) a CVP (mmhg) Heart rate (min 1 ) Qp/Qs a Fio 2 (%) a 0.06 Ventilation index PH a Pco 2 (mm Hg) Po 2 (mm Hg) Base excess (meq/l) a SVRI Total PVRI a Difference from alive without complications. b Difference from death. AVo 2 arteriovenous oxygen content difference; CCP continuous cerebral perfusion; CVP central venous pressure; Fio 2 fraction of inspired oxygen; MAP mean arterial pressure; Pco 2 partial pressure of carbon dioxide; Po 2 partial pressure of oxygen, PVRI peripheral vascular resistance index; Qp/Qs pulmonary-to-systemic flow ratio; Sao 2 arterial oxygen saturation; Svo 2 superior vena cava oxygen saturation; SVRI systemic vascular resistance index. outcome model by younger age, lower weight, and longer duration of intraoperative support, lower Svo 2, higher Sao 2, lower heart rate, and lower ph. For the purposes of controlling for changes in outcome over time, the cohort was divided into three equal eras. There was no significant difference in outcome end points of uncomplicated survival, survival with complications, or early death among the three eras (p 0.357). Figure 3 shows the hourly Svo 2 for all 116 patients separated by the outcome end points of uncomplicated survival, survival with complications, and early death. The Svo 2 level was higher in uncomplicated survivors compared with survivors with complications, and in turn, survivors with complications had higher Svo 2 levels than patients who died early (p 0.03). Survivors also showed a gradual increase in Svo 2, whereas patients in the early mortality group had a decrease in Svo 2 in the first 6 hours after return to the intensive care unit. In an attempt to determine the potential for Svo 2 monitoring to improve outcome, we looked at the three outcome end points among patients with the lowest Svo 2 after arrival to the intensive care unit (Fig 4). Among the 29 patients in the lowest quartile, the initial Svo 2 level on arrival in the intensive care unit was not different with respect to the three outcome end points of uncomplicated survival, survival with complications, and early death.

6 1306 TWEDDELL ET AL Ann Thorac Surg OXIMETRY MONITORING AFTER STAGE 1 PALLIATION 2007;84: Table 4. Multivariable Analysis of Impact of Factors on Outcome Factor Complications R Death R Ord log model R Age (younger) Weight (lower) Gestational age Aortic diameter Phenoxybenzamine Additional diagnosis DHCA duration Total support time (longer) SvO 2 lower SaO 2 higher AVO 2 Hemoglobin (higher) MAP CVP Heart rate (lower) Qp/Qs (lower) Fio 2 (higher) ph (lower) Base Excess Pco 2 Po 2 AVo 2 arteriovenous oxygen content difference; CVP central venous pressure; DHCA deep hypothermic circulatory arrest; Fio 2 fraction of inspired oxygen; MAP mean arterial pressure; Pco 2 partial pressure of carbon dioxide; Po 2 partial pressure of oxygen, PVRI peripheral vascular resistance index; Qp/Qs pulmonary-to-systemic flow ratio; Sao 2 arterial oxygen saturation; Svo 2 superior vena cava oxygen saturation. Fig 3. The hourly superior vena cava oxygen saturation (Svo 2 ) for all 116 patients is shown with the standard deviation (error bars), separated by the outcome endpoints of uncomplicated survival, survival with complications, and early mortality. The Svo 2 was higher in uncomplicated survivors compared with survivors with complications, and in turn, survivors with complications had higher Svo 2 than patients who sustained early mortality (p 0.03). In addition survivors, showed a gradual increase in Svo 2, whereas patients in the early mortality group had a decrease in SvO 2 in the first 6 hours after return to the intensive care unit. the relative risk of ECMO, CPR, and mortality rate for patients with an Svo 2 that was more or less than 55%, the 25th percentile for average 48 hour postoperative Svo 2. During the first 18 hours, however, there was a difference in Svo 2 among these patents such that patients in the uncomplicated survival and complicated survival groups showed a gradual increase in Svo 2, whereas those in the early death group showed no substantial improvement in Svo 2 (p 0.04). A separate analysis was undertaken to determine the clinical utility of Svo 2 monitoring in predicting the end points of ECMO, CPR, and death. Patients who required CPR not related to shunt occlusion had a borderline lower Svo 2 (54.1% 9.2% versus 59.2% 9.2%, p 0.063) and a significantly wider AVo 2 ( ml/dl versus ml/dl, p 0.014). Use of ECMO was associated with a lower Svo 2 (54.0% 9.7% versus 59.9% 9.2%, p 0.031) and wider AVo 2 ( ml/dl versus ml/dl, p 0.014). Patients who died after S1P had a lower Svo 2 (53.1% 10.6% versus 59.3% 9.2%, p 0.034) and wider AVo 2 ( ml/dl versus ml/dl, p 0.035). The risk of outcome end points at incremental ranges of Svo 2 is illustrated in Figure 5. The Svo 2 was inversely related to the risk of any complication, CPR, ECMO at any time, and early and late death. Table 5 summarizes Fig 4. Among patients in the lowest 25th percentile for superior vena cava oxygen saturation (Svo 2 ) upon arrival to the cardiac intensive care unit, a failure of the Svo 2 to normalize in the first 18 hours was characteristic of those in the early mortality group (black circles). These data suggest that efforts to increase a low SvO 2 are successful in a proportion of patients with the lowest Svo 2. (Error bars show the standard deviation. Clear circles uncomplicated survival; patterned circles survival with complications.)

7 Ann Thorac Surg TWEDDELL ET AL 2007;84: OXIMETRY MONITORING AFTER STAGE 1 PALLIATION 1307 Fig 5. Risk of complications according to postoperative superior vena cava oxygen saturation (Svo 2 ) assessed hourly for 48 hours. The * indicates a significant difference from risk at lower Svo 2 in time-series regression. (Error bars show the standard deviation. Blue line any complication; black line any mortality; gray line cardiopulmonary resuscitation [CPR]; orange line extracorporeal membrane oxygenation [ECMO]; green line early death; red line early ECMO.) An Svo 2 of less than 55% was associated with the need for ECMO during the first 48 postoperative hours, the use of any ECMO during S1P hospitalization, CPR, and early death. Comment The early postoperative period after the Norwood procedure is one of high risk due to the compounding of several physiologic vulnerabilities. The single right ventricle is an inferior power source further compromised by a recent period of ischemia and CPB. With limited cardiac output, precise partitioning of the pulmonary and systemic flow is necessary to achieve adequate tissue oxygen delivery. The combination of decreased cardiac output and the inherent inefficiencies of parallel circulation is the hallmark of the early postoperative period after the Norwood procedure. Maintenance of adequate systemic oxygen delivery is the goal. Monitoring during this period has historically included blood pressure, heart rate, CVP, and measurement of systemic arterial oxygen saturation using pulse oximetry. Clinical assessment of perfusion is used despite data suggesting that clinical assessment of cardiac output, even by trained clinicians, is often inaccurate [6]. Objective measurement of systemic oxygen delivery would permit better management of the patient during this vulnerable period after the Norwood procedure. For the last 10 years, we have used venous oximetry as an objective assessment of systemic oxygen delivery and to guide management. The goal of postoperative management is to achieve oxygen delivery adequate to meet the metabolic demands of the tissues and enable healing. Physical examination is limited in its ability to accurately assess cardiac output, despite defining precise areas within the physical examination such as pulse amplitude, extremity temperature, and capillary refill [6 8]. Serial measurements of lactate have been used as a guide for the postoperative management of patients with HLHS. Decreasing lactate level was associated with an uncomplicated postoperative course and a rising level was shown to predict the need for ECMO [9]. Unfavorable changes in lactate would indicate that a period of oxygen deliverydependent oxygen consumption is already in effect. Therefore, measurement of lactate does not predict a period of anaerobic metabolism but rather demonstrates that anaerobic metabolism is already in place. Assumptions concerning cardiac output based on the commonly measured parameters of arterial saturation and blood pressure are inaccurate [10, 11]. The mixed venous blood is a summary of the last blood in contact with the tissues at the capillary level and can be used as a guide to the tissue oxygen economy. Superior vena cava saturation is a close approximation of mixed venous saturation and has been validated as a target for intervention that will result in improved outcome. In a study to look at the impact of Svo 2 monitoring and shock, Rivers and colleagues [12] randomized patients to receive conventional monitoring (CVP, blood pressure, and urine output) or early goal-directed therapy that included Svo 2 monitoring. They showed a re- Table 5. Relative Risk of Complications With Superior Vena Cava Oxygen Saturation at More or Less Than 55% Outcome Svo 2 55% Svo 2 55% Relative Risk p Value Early ECMO (48 hour) 9% 0% Any ECMO 21% 4% CPR 15% 4% Mortality (30 day) 15% 3% Mortality (any) 29% 15% CPR cardiopulmonary resuscitation; ECMO extracorporeal membrane oxygenation; Svo 2 superior vena cava oxygen saturation.

8 1308 TWEDDELL ET AL Ann Thorac Surg OXIMETRY MONITORING AFTER STAGE 1 PALLIATION 2007;84: duction in organ failure as well as early and late mortality in those patients in whom real-time oxygen delivery was assessed objectively and used in treatment. Intermittent sampling of the superior vena cava blood has been used to assist with postoperative management of patients undergoing palliation for HLHS. Rossi and colleagues [13] used intermittent SVC blood samples to measure venous saturation and calculate both Qp/Qs and AVo 2. Survivors showed a gradual increase in venous saturation and a gradual decline in Qp/Qs during the first 24 hours. They noted that Qp/Qs could be elevated even in patients with an acceptable arterial saturation [13]. We reported the use of continuous superior vena cava saturation monitoring using a 4F oximetric catheter. Of importance was that we identified abrupt decreases in Svo 2, indicating a sudden decrease in systemic oxygen delivery that could not be identified by monitoring Sao 2, MAP, heart rate, or CVP. These sudden decreases in Svo 2 were mirrored by subtle increases in MAP and Sao 2 and represent acute increases in SVR. These observations led us to incorporate phenoxybenzamine into our routine management for patients undergoing SIP. In that study we demonstrated that continuous Svo 2 could identify patients with decreasing systemic oxygen delivery and that interventions could restore the venous saturation to an acceptable range [1]. In subsequent work we showed that subsets of patients with HLHS who have been considered high risk had decreased Svo 2 in the early postoperative period, suggesting that the increased risk was partly due to lower systemic oxygen delivery [14]. Although that study included a small number of patients, the low mortality suggested that careful attention to systemic oxygen delivery by use of continuous Svo 2 monitoring could improve the outcome of high-risk subtypes of HLHS. In an analysis of a broad group of HLHS and variants, we found that use of continuous Svo 2 favored survival of patients undergoing S1P [4]. Further work from our group showed an inverse correlation between Svo 2 and the development of metabolic acidosis. Although we have targeted an Svo 2 of 50%, we failed to achieve this target in many patients, especially in the early postoperative period. Analysis of the relationship between Svo 2 and increasing metabolic acidosis revealed an anaerobic threshold when Svo 2 fell towards 30% [15]. The current study suggests that targeting an even higher threshold for Svo 2, in the 55% to 60% range, is optimal to achieve the lowest risk of death and complications. Li and colleagues [11] have shown that oxygen consumption is highly variable in the postoperative patient and that improvement in the postoperative oxygen economy can be achieved through strategies that limit oxygen consumption, such as control of hyperthermia and sedation. Fundamental management of patients after S1P now includes not only the usual strategies of optimizing rhythm, rate, preload, and contractility but also the routine use of sustained afterload reduction, a Fio 2 high enough to prevent pulmonary venous desaturation, red cell transfusion to maximize oxygen carrying capacity, and efforts to minimize oxygen consumption [3, 16, 17]. Continuous monitoring of oxygen economy has allowed the identification of shock states and has provided a real-time guide to the efficacy of intervention. In this group of patients with the most agreed upon anatomic definition of HLHS, Svo 2 was shown to identify those at risk for mortality and complications. The early mortality in this group was 7% and compares favorably with any contemporary series. Commonly measured hemodynamic parameters such as blood pressure were not predictive of outcome. Other commonly cited risk factors such as the presence of additional diagnoses, aortic atresia versus aortic stenosis, low birth weight, and prematurity were not identified as risk factors for mortality [18, 19]. In this series, no patient with low birth weight, prematurity, or additional extracardiac diagnosis died. These findings suggest that Svo 2 monitoring may have ameliorated the impact of these commonly identified risk factors for death. The incidence of CPR and ECMO was similar to other recent series of patients undergoing S1P of HLHS. The need for CPR and ECMO as rescue therapy for cardiac failure has been eliminated since the year 2000 and reflects an increasing reliance on Svo 2 to guide management of low cardiac output syndrome. It is important to note that Svo 2 monitoring has not eliminated the need for CPR or ECMO due to acute shunt occlusion. The incidence of noncardiac complications was low in this series. There was a single incident of stroke in this cohort. Clinically evident seizures occurred in 5 patients (4.3%), which compares favorably with the 18% incidence of seizures reported by Clancy and colleagues [20]. Our own experience relating Svo 2 with late neurodevelopment outcome indicates that the potential for neurologic injury is inversely related to Svo 2 [21]. The incidence of renal failure was low in this series, with only 2 patients (1.7%) requiring dialysis. Three patients (2.5%) had renal insufficiency, defined as oliguria with hyperkalemia, requiring therapy. This incidence of renal failure compares favorably with the incidence of renal failure in other series of patients with congenital heart disease, including those with less severe forms [22, 23]. Our series was unique for a very low incidence of gastrointestinal complications. Only 1 patient in our series had necrotizing enterocolitis, and this patient did not require surgical intervention. A higher incidence of gastrointestinal complications has been reported by other experienced centers. McElhinney and colleagues [24] reported a 7.6% incidence of necrotizing enterocolitis in patients with HLHS, and Jeffries and colleagues [25] reported an 18% incidence. There is no dispute that decreased cardiac output is a risk factor for both renal insufficiency and necrotizing enterocolitis. Continuous Svo 2 monitoring has shortcomings. It is invasive, with potential complications of bleeding and vessel thrombosis. We have had one episode of bleeding requiring reexploration but no episodes of thrombosis. To minimize the risk of thrombosis, we avoid placing any additional catheters in the superior vena cava.

9 Ann Thorac Surg TWEDDELL ET AL 2007;84: OXIMETRY MONITORING AFTER STAGE 1 PALLIATION 1309 In cases of anomalous pulmonary venous drainage to the superior vena cava, saturations will not reflect the tissue oxygen economy. In cases of distributive shock, venous saturations may appear satisfactory but oxygen delivery may be maldistributed at the tissue level. Continuous Svo 2 monitoring is of little additional benefit to conventional monitoring in identifying shunt thrombosis. The abrupt decrease in arterial saturation and change in the shunt murmur are generally diagnostic. Likewise, lethal arrhythmias are seen in the postoperative Norwood patient and can be diagnosed by continuous electrocardiograph monitoring and hypotension. Nonetheless, if a prodrome of decreased cardiac output preceded the lethal arrhythmias, Svo 2 monitoring may identify patients at risk. Mixed venous saturation monitoring may be relatively insensitive to significant regional hypoperfusion, which can occur during early shock states, for which gastric tonometry or near infrared spectroscopy may provide better information. All of the patients in this study had Svo 2 monitoring, and no control group was used to prove the efficacy of Svo 2 monitoring. We did not quantify our ability to increase a low Svo 2. We used assumed values for both pulmonary venous saturation and oxygen consumption. We rarely used a Fio 2 of less than 30%, and therefore, it is unlikely that pulmonary venous desaturation complicated our calculations. Although it has been shown that oxygen consumption can vary greatly in the early postoperative period, the major findings of this study are related to Svo 2 and AVo 2 and therefore did not require any assumptions. A large proportion of the data in this study were prospectively collected; however, data concerning many of the complications were retrospectively collected. The complications selected were those used by other authors and were fairly easy to identify by electronic database review methods. It would unlikely to miss any patients who required dialysis or needed an operation for necrotizing enterocolitis. Goal-directed therapy with Svo 2 as an indicator of systemic oxygen delivery is associated with excellent early survival and a low incidence of organ failure in infants after S1P for HLHS. Inability to optimize Svo 2 in the early postoperative period is associated with an increased risk of organ failure, ECMO, and death. References 1. Tweddell JS, Hoffman GM, Fedderly RT, et al. Phenoxybenzamine improves systemic oxygen delivery after the Norwood procedure. Ann Thorac Surg 1999;67: Tchervenkov CI, Jacobs JP, Weinberg PM, et al. The nomenclature, definition and classification of hypoplastic left heart syndrome. Cardiol Young 2006;16: Tweddell JS, Hoffman GM. Postoperative management in patients with complex congenital heart disease. Semin Thorac Cardiovasc Surg Pediatr Card Surg Annu 2002;5: Tweddell JS, Hoffman GM, Mussatto KA, et al. Improved survival of patients undergoing palliation of hypoplastic left heart syndrome: lessons learned from 115 consecutive patients. Circulation 2002;106(suppl I):I Tweddell JS. The Norwood procedure with an innominate artery-to-pulmonary artery shunt. Operat Tech Thorac Cardiovasc Surg 2005;7: Tibby SM, Hatherill M, Marsh MJ, Murdoch IA. Clinicians abilities to estimate cardiac index in infants and children. Arch Dis Child 1997;77: Leonard PA, Beattie TF. Is measurement of capillary refill time useful as part of the initial assessment of children? Eur J Emerg Med 2004;11: Tibby SM, Hatherill M, Murdoch IA. Capillary refill and core-peripheral temperature gap as indicators of hemodynamic status in pediatric intensive care patients. Arch Dis Child 1999;80: Charpie JR, Dekeon MK, Goldberg CS, Mosca RS, Bove EL, Kulik TJ. Serial lactate measurements predict early outcome after neonatal repair or palliation for complex congenital heart surgery. J Thorac Cardiovasc Surg 2000;120: Taeed R, Schwartz SM, Pearl JM, et al. Unrecognized pulmonary venous desaturation early after Norwood palliation confounds Qp:Qs assessment and compromises oxygen delivery. Circulation 2001;103: Li J, Zhang G, Holtby HM, et al. Inclusion of oxygen consumption improves the accuracy of arterial and venous oxygen saturation interpretation after the Norwood procedure. J Thorac Cardiovasc Surg 2006;131: Rivers E, Nguyen B, Havstad S, et al. Early goal-directed therapy in the treatment of severe sepsis and septic shock. N Engl J Med 2001;345: Rossi AF, Sommer RJ, Lotvin A, et al. Usefulness of intermittent monitoring of mixed venous oxygen saturation after stage I palliation for hypoplastic left heart syndrome. Am J Cardiol. 1994;73: Tweddell JS, Hoffman GM, Fedderly RT, et al Patients at risk for low systemic oxygen delivery after the Norwood procedure. Ann Thorac Surg 2000;69: Hoffman GM, Ghanayem NS, Kampine JM, et al. Venous saturation and the anaerobic threshold in neonates after the Norwood procedure for hypoplastic left heart syndrome. Ann Thorac Surg 2000;70: Hoffman GM, Tweddell JS, Ghanayem NS, et al Alteration of the critical arteriovenous oxygen saturation relationship by sustained afterload reduction after the Norwood procedure. J Thorac Cardiovasc Surg 2004;127: Bradley SM, Atz AM, Simsic JM. Redefining the impact of oxygen and hyperventilation after the Norwood procedure. J Thorac Cardiovasc Surg. 2004;127: Stasik CN, Gelehrter S, Goldberg CS, Bove EL, Devaney EJ, Ohye RG. Current outcomes and risk factors for the Norwood procedure. J Thorac Cardiovasc Surg 2006;131: Gaynor JW, Mahle WT, Cohen MI, et al. Risk factors for mortality after the Norwood procedure. Eur J Cardiothorac Surg 2002;22: Clancy RR, McGaurn SA, Goin JE, et al. Allopurinol neurocardiac protection trial in infants undergoing heart surgery using deep hypothermic circulatory arrest. Pediatrics 2001; 108: Hoffman GM, Mussatto KA, Brosig CL, et al. Systemic venous oxygen saturation after the Norwood procedure and childhood neurodevelopmental outcome. J Thorac Cardiovasc Surg. 2005;130: Chan K, Ip P, Chiu C, Cheung YF. Peritoneal dialysis after surgery for congenital heart disease in infants and children. Ann Thorac Surg 2003;76: Dittrich S, Dahnert I, Vogel M, et al. Peritoneal dialysis after infant open heart surgery. Ann Thorac Surg 1999;68: McElhinney DB, Hedrick HL, Bush DM, et al. Necrotizing enterocolitis in neonates with congenital heart disease: risk factors and outcomes. Pediatrics 2000;106: Jeffries HE, Wells WJ, Starnes VA, Wetzel RC, Moromisato DY. Gastrointestinal morbidity after palliation for hypoplastic left heart syndrome. Ann Thorac Surg 2006;81:982 7.

10 1310 TWEDDELL ET AL Ann Thorac Surg OXIMETRY MONITORING AFTER STAGE 1 PALLIATION 2007;84: Appendix List of Variables Evaluated Demographics Age Weight Gestational at birth Date of operation Sex Anatomic subtype Aortic atresia/mitral atresia Aortic atresia/patent mitral valve Aortic stenosis/patent mitral valve Aortic stenosis/patent mitral valve/ventricular septal defect Ascending aortic diameter Additional diagnoses Cardiac Noncardiac Preoperative factors Antenatal diagnosis Ventilation Inotropic support Operative variables Total support time Continuous cerebral perfusion Phenoxybenzamine Duration of deep hypothermic circulatory arrest Shunt size Postoperative variables Time to chest closure Hospital length of stay Hospital survival Cardiopulmonary resuscitation Extracorporeal membrane oxygenation Organ failure Circulatory Arrhythmia Congestive heart failure Extracorporeal membrane oxygenation Cardiopulmonary resuscitation Renal failure requiring dialysis Gastrointestinal Necrotizing enterocolitis Ischemic hepatitis Neurologic Seizures (clinical) Intraventricular hemorrhage Stroke Hemodynamic and other variables during first 48 hours Heart rate Mean arterial pressure Central venous pressure Arterial saturation Superior vena cava saturation Arteriovenous oxygen content difference Pulmonary-to-systemic blood flow ratio Hemoglobin Pressure of carbon dioxide ph Base deficit Fraction of inspired oxygen Ventilation index DISCUSSION DR JEFFREY P. JACOBS (St. Petersburg, FL): Thanks, Jim. Those clearly are among the best results for stage I palliation in the world, and that is a very impressive series. You and I have talked about this before about the role of near-infrared spectroscopy, which I guess one could view as a complimentary modality to Svo 2. Do you think that using the NIRS technology would ever replace putting a catheter in the SVC, or would you think that it would just be something where you would continue to do both? DR TWEDDELL: Near-infrared spectroscopy is an excellent adjunct to Svo 2 monitoring. It is a terrific trend monitor. In addition, near infrared spectroscopy permits identification of changes in regional perfusion. We place one probe on the forehead and one probe on the back at around the L2 level. In this way, we can monitor changes in the relative distribution of cardiac output to the cerebral and splanchnic circulations. I think you need to combine NIRS monitoring with venous saturation monitoring. Will it eventually take over? I think that is possible. Certainly it would be nice to have a noninvasive device and something you could use more long term as well. DR JACOBS: But right now you still put the SVC catheter in all the stage 1s? DR TWEDDELL: Yes. DR CARL L. BACKER (Chicago, IL): Jim, that is a great series, and I really think you may be changing the standard of care for postoperative follow-up after the Norwood operation. I have heard Nancy Ghanayem, one of your coauthors, talk about this, and she made the statement, and I think I heard her correctly, that she would prefer to have the NIRS monitoring than an arterial line. Do you agree with that or is she out on a limb? DR TWEDDELL: The NIRS monitoring does provide some very useful information. It is best as a trend monitor, and we use it more and more. If we have any concerns about a patient, we will place the probes on the patient and get an idea of what is going on. Some of the changes that occur as a result of abrupt events are pretty dramatic and are picked up by NIRS instantaneously. It is a powerful tool. Exactly how it is going to be used and so on is a little difficult to predict right now. To answer your question, I want the A-line, I want the Svo2 monitor, I want the whole thing because the more objective data you are able to gather, the less likely you are to be fooled. If we had all the information, we d all be more likely to make the same

11 Ann Thorac Surg TWEDDELL ET AL 2007;84: OXIMETRY MONITORING AFTER STAGE 1 PALLIATION 1311 correct decisions in a more timely manner and we would all achieve the best possible outcomes. DR BACKER: The other question I have is given all this data, I mean, your reliance on the Svo 2, when you see that it is not coming up or is going down, are you now, in the last quartile of your study going to ECMO much sooner than you were earlier in the series? Also, what is going through your mind when this happens? What kind of changes are you making when you see that Svo 2 going down into the basement? DR TWEDDELL: That is an excellent question. In addition to our studies, we paid a lot of attention to some of the terrific work that Dr Li has done in Toronto and the work that was done in Cincinnati by Taeed and colleagues. In addition to the usual measures of optimizing rate, rhythm, filling pressures, inotropy, afterload reduction is absolutely essential. It doesn t have to be phenoxybenzamine, but you should be using something such as milrinone to control the SVR. You want to use an Fio 2 that will prevent pulmonary venous desaturation, ventilating a very low Fio 2 is a mistake. You want to augment the oxygen carrying capacity by increasing the hematocrit. We push the hematocrit up to at least 45% in the uncomplicated patient, and in the sicker patient, an hematocrit over 50% may be necessary. You need to reduce oxygen consumption if you can. You want to avoid hyperthermia, use sedation and paralysis to decrease oxygen consumption, and try to match oxygen delivery and consumption. But if all that doesn t work, you should go on ECMO. I would prefer to go on ECMO before chest compressions are initiated to limit neurologic injury. DR BACKER: One final question. What about stress dose steroids? Are you giving that to some of the neonates? All of the neonates? Just the ones that have a low Svo 2? DR TWEDDELL: We give preoperative steroid Solu-Medrol to everybody, and we give Solu-Medrol at midnight and 6 am the day before surgery. DR BACKER: But what about post-op? DR TWEDDELL: On a per-patient basis. Just one more point about the low Svo 2, George Hoffman published a paper about a year ago looking at late neurodevelopmental outcome and relating it to this perioperative database, and one of his findings was that a low Svo 2 in the early postoperative period, not surprisingly, predicted a poor neurodevelopmental outcome. So that may be another reason as Ross Ungerleider has suggested to use postcardiotomy support not simply to prevent mortality but to limit neurologic morbidity. DR BACKER: It s a great contribution. DR CHRISTOPHER A. CALDARONE (Toronto, Ontario, Canada): Jim, just one question. We can t challenge your results, of course, but can we challenge one of your conclusions today? About 10% of your patients were on ECMO. Your program is well known for having a very standardized approach to patient management. If threshold Svo 2 is used to trigger putting a patient on ECMO, then is it fair to make the conclusion that low Svo 2 is associated with complications because you said ECMO is listed as one of your complications, as an end point of a complication. It is kind of self-fulfilling if low Svo 2 s result in ECMO, then it is going to look like low Svo 2 results in the basket term of complications which includes ECMO. So my question is, if you took ECMO out of the analysis as an end point as a complication, does low Svo 2 predict all those other complications that you saw? DR TWEDDELL: We tried to pick complications that were related to low output. I understand your point, but I think ECMO is a marker for inadequate cardiac output certainly. DR MUHAMMAD A. MUMTAZ (Cleveland, OH): Just a very brief question. I may have missed that on your slides, but when the Svo 2 was where you desired it to be, about 45, how often did you have complications? I mean, is that a time when you can just go home and sleep and say things are nice? DR TWEDDELL: We didn t present the data here, but it is in the manuscript we dichotomized at about 55% to see how that did with predicting outcome, and it certainly predicted the use of ECMO/CPR and mortality. But if you really wanted to avoid complications, have complication-free survival, you needed a higher Svo 2, really greater than 60%. So it suggests that at least at the tissue oxygen economy level, the single-ventricle patient may not have any more tolerance for low output and a decreased oxygen delivery than any other baby. Thank you.