Optimal pulmonary to systemic blood flow ratio for best hemodynamic status and outcome early after Norwood operation

Surgical modifications and management improvements have led to increased survival after first-stage Norwood palliation of the hypoplastic left heart syndrome (HLHS). However, mortality still exceeds that of most reconstructive procedures for complex congenital heart defects. It is generally assumed, that pulmonary to systemic blood flow ratio (Qp/Qs) close to 1 is associated with the best clinical presentation for neonates with HLHS [1,2]. As pulmonary vascular resistance can change, sometimes widely, after completion of stage I palliation, Qp/Qs will also vary, despite adequate management. Imbalances of Qp/Qs and limited myocardial reserve of the morphological right ventricle compounded with inadequate systemic oxygen delivery and end-organ perfusion account for most of early mortality. To optimize systemic oxygen delivery we, like some centers [3,4], have adopted postoperative continuous monitoring of systemic venous oxygen saturations (SvO2) to allow online Qp/Qs measurements. A mathematical model developed by Barnea et al. [1] suggests that maximal oxygen delivery occurs at Qp/Qs of less than 1. However, it remains unclear, whether this model can be transferred to the postoperative clinical situation. It is also not obvious which optimal Qp/Qs should be strived after. The objective of this study was to evaluate the level of Qp/Qs that would correlate with best postoperative hemodynamic status, and thus, better outcome. Hemodynamic data of 25 consecutive patients, who underwent modified Norwood operation for HLHS or for complex forms of single ventricle with systemic outflow obstruction between October 2002 and January 2005, were analyzed in retrospect. One patient, operated upon during the period, who could not be separated from cardiopulmonary bypass after completion of Norwood reconstruction, was excluded. Age, weight, diagnosis, and preoperative clinical condition of all these 26 patients are displayed in Table 1. Patient demographics Details of perioperative management and operative conduct have been recently reported [5]. In particular, surgery included aortic arch augmentation using pulmonary homograft material, atrial septectomy, and placement of a polytetrafluoroethylene shunt either from the innominate artery (n = 19, diameter 3.5–5 mm) or from the right ventricle (n = 7, diameter 5 or 6 mm) to the pulmonary artery. The aortic arch was reconstructed with continuous antegrade cerebral perfusion in all patients and the Norwood procedure was performed on a beating heart in 13 patients [6]. Oximetric catheters (4F Edwards Life Sciences, Irvine, CA, USA) were placed through the right atrium into the superior vena cava to allow continuous monitoring of systemic venous oxygen saturation (SvO2). An additional line was placed in the common atrium for pressures monitoring and infusion of inotropic drugs. Sternum was routinely left open usually for 2 days until hemodynamic stable conditions were achieved. Table 2 displays management targets, according to our protocol. It is to be noted that we aimed at an SvO2 of about 50% and a systemic arterial oxygen saturation (SaO2) between 75 and 80%. This would correlate with an arteriovenous oxygen saturation difference (AVD = SaO2 − SvO2) of about 25%, and a pulmonary to systemic blood flow ratio (Qp/Qs) of about 1. Qp/Qs was estimated according to the Fick method: (SaO2 − SvO2)/(SaO2 − SpvO2), with SpvO2: pulmonary venous oxygen saturation, assumed to be 97%. Oxygen excess factor, which has been shown to correlate with systemic oxygen delivery [4], was calculated as SaO2/AVD. Indexed pulmonary blood flow (Qp = VO2/[0.136 times hemoglobin(g/dl) × {97% − SaO2%}]), indexed systemic blood flow (Qs = VO2/[0.136 times hemoglobin(g/dl) × {SaO2 − SvO2%}]), cardiac index (CI = Qp + Qs), and systemic vascular resistances were calculated using oxygen consumption (VO2) derived from standard formulae [7]. Dissolved oxygen content was assumed to be negligible. All patients received dopamine 4–6 μg/(kg min), milrinone 0.5 μg/(kg min), and phentolamine 2–8 μg/(kg min). Norepinephrine or epinephrine was added if supplementary inotropic support became necessary to achieve postoperative management targets (Table 2). The dosage of milrinone and/or phentolamine was adapted in case of acidosis, increasing lactate levels or low urinary output, in order to reduce systemic afterload and increase systemic oxygen delivery, even if Qp/Qs < 1 ensued. Patients received adequate sedation using a continuous infusion of fentanyl (5–10 μg/(kg min)) and midazolam (1–4 μg/(kg min)) until chest closure. Ventilator settings were adjusted to maintain normocapnia (pCO2 40–50 Torr) with lowest possible oxygen concentration (FiO2) to achieve the aimed arterial and venous oxygen saturation. The addition of CO2 to the inspired gas was not used. End-expiratory pressure was routinely kept at 8–10 cmH2O until chest closure, anticipating pulmonary venous desaturation from ventilation/perfusion mismatch. Postoperative management targets A prospective perioperative database for all Norwood patients was established including demographic, surgical, and postoperative hemodynamic and laboratory data. Informed consent was obtained from patients’ parents for data collection and statistical analysis for study purposes, in accordance to national laws. All hemodynamic parameters, blood gas analyses, serum lactate levels, urinary output, and dosages of inotropes and furosemide used were entered 3 hourly into the database from admission to intensive care to the first 48 postoperative hours, or until the patient expired or extracorporal membrane oxygenation (ECMO) was initiated. Mathematical model implies no linear relationship between Qp/Qs and systemic oxygen delivery. As Qp/Qs increases, systemic oxygen availability initially increases and reaches a maximum at a level <1, then slowly decreases until it reaches a critical value of about 2, when it drops rapidly [1]. To validate this theoretical model and the findings derived from it, clinically, after stage I palliation, all hemodynamic data samples were pooled, irrespective of preoperative and operative variables, thus, irrespective of patients. Data were afterwards assigned to three groups according to their corresponding Qp/Qs: Group 1: calculated Qp/Qs ≤ 1, Group 2: calculated Qp/Qs between 1 and 2, Group 3: calculated Qp/Qs ≥ 2. In that way, data from any patient collected across the whole study period could be assigned to any group according to calculated Qp/Qs, at each specific time point. After grouping, the corresponding hemodynamic and oximetric parameters were compared. Moreover, Qp/Qs ratios in the first 48 h as well as lactate levels were compared in hospital survivors and non-survivors. Data were summarized as mean ± SEM. Independent Student's t-test was used to compare means of each hemodynamics variable for significant differences in between groups. Levene's test was employed to test for equality of variances. Fisher's exact test or the χ2-test was used, as appropriate. Differences were considered statistically significant at a p-value of <0.05. Analyses were performed using the statistical software package SPSS 11.0 (SPSS Inc., Chicago, IL, USA). A total of 343 Qp/Qs samples could be collected from the 25 patients comprising 1095 patient-hours (mean 43.6 ± 2.3 h/patient) 80% (343/425) of completely collectable hemodynamic data. The unavailability of data was due either to death within 48 h (n = 3) or to the use of extracorporeal membrane oxygenation (ECMO, n = 1). One hundred and ten samples were assigned to group 1 with a mean Qp/Qs of 0.7 ± 0.02, 184 to group 2 (mean Qp/Qs 1.4 ± 0.02), and 49 to group 3 (mean Qp/Qs 2.5 ± 0.08). The three groups were significantly different: p < 0.001. According to the standardized management protocol, hemodynamic targets such as mean arterial pressure (MAP) of about 50 mmHg, SvO2 above 50% and SaO2 between 70 and 80% were reached in all groups, despite significantly different Qp/Qs. Differences with respect to hemodynamic variables, oximetric data, and computed cardiac indices are displayed in Table 3. MAP and SVR were lowest in group 1, resulting in increased systemic oxygen delivery, indicated by lower AVD (p < 0.001) and elevated oxygen excess factor (p < 0.001). However, common atrial pressure (p = 0.02), norepinephrine doses (p = 0.006), and lactate levels (p = 0.009) were higher, and urinary output (p = 0.005) lower, compared to group 2, indicating worse hemodynamic status. Imbalanced circulations noted in groups 1 and 3 were accompanied by higher calculated total cardiac index (p = 0.02 and p = 0.01) and lower urinary output (p = 0.005 and p = 0.08), when compared to group 2. Hemodynamic variables Four among the 25 studied patients died: overall hospital mortality of 19% (5/26) including one operative death. Death occurred at 8, 14, 46, and 60 h after admission to Intensive Care Unit. In one patient, ECMO had been initiated 18 h after surgery but was discontinued, because of cerebral bleeding after 42 h of support. Preoperative, operative, and postoperative risk factors for hospital mortality are displayed in Table 4. Univariate risk factor analysis for hospital mortality Qp/Qs ratios and courses of serum lactate levels in survivors and non-survivors are displayed in Figs. 1 and 2. Qp/Qs remained above 1 in survivors and was, most of the time, below 1 in the four patients who died. Lactate levels were under 4 mmol/l and decreasing in survivors. They were, most of the time, above 4 mmol/l or increasing in non-survivors. Qp/Qs ratio in survivors and non-survivors: calculated Qp/Qs ratio was higher in survivors than in non-survivors. Lactate levels in survivors and non-survivors: lactate levels were lower than 4 mmol/l or decreasing in survivors, on the contrary they were higher than 4 mmol/l or increasing in non-survivors, most of the times. This observational study describes three levels of Qp/Qs (≤1, 1–2, and ≥2) with corresponding hemodynamic and oximetric parameters and derived calculated indices, at each time point the sample was collected. The surveillance phase was limited to the first 48 h after stage I palliation, since the majority of all deaths occurs within this interval [8,9]. Improvements of outcome within this time period, therefore, should result in better overall outcome for patients undergoing Norwood reconstruction. By pooling of all samples across the study period, homogeneity of patients following a standardized perioperative management protocol was postulated. Interventions were not randomized or blinded. Instead, management strategy was based on assumptions derived from theoretical models. Pulmonary and systemic vascular resistances were modified to achieve the assumed optimal Qp/Qs of 1 and SvO2 of above 50%. This strategy is reflected by the different group sizes, with the smallest number of samples in group 3. Qp/Qs calculations are based on the assumption of pulmonary venous saturation being 97%. However, pulmonary venous desaturation due to pulmonary disease, ventilation perfusion mismatch, or to pulmonary arteriovenous fistulae occurs frequently after stage I palliation, resulting in underestimation of Qp/Qs. Unfortunately, placement of a catheter into a pulmonary vein to measure pulmonary venous saturations in addition to an oximetric catheter and another one for direct measurement of common atrial pressure, seems unrealistic and may increase risk of thrombembolism to an unacceptable level. Anticipating pulmonary venous desaturation, we generally keep an increased end-expiratory pressure of 8–10 cmH2O until chest closure as suggested by Taeed et al. [10] in order to prevent pulmonary venous desaturation from ventilation perfusion mismatch with pulmonary edema or atelectasis. Despite these limitations, the actual study helps to reduce errors arising from assumptions of optimal Qp/Qs derived from mathematical models. It provides clinical evidence by linking Qp/Qs with hemodynamic parameters and outcome. Clinically, Qp/Qs between 1 and 2 correlated with best postoperative hemodynamic status and consequently better outcome and lower mortality. Most samples were assigned to this group and the majority of these patients had unremarkable recovery, with higher urinary output, lower lactate levels and cardiac indices, indicating less volume load for the single ventricle. With consequent systemic afterload reduction therapy, pulmonary overcirculation with Qp/Qs ≥ 2 was uncommon (14%) and did not result in early mortality in our cohort. Hemodynamic status was not much different from group 2 samples, except for a higher MAP with corresponding lower doses of norepinephrine needed to accomplish our standardized hemodynamic targets. However, pulmonary overcirculation was associated with higher cardiac index compared to group 2. Common atrial pressure and urinary output in group 3 were not significantly different from group 1, indicating that volume load and cardiac performance is worse than in group 2 samples. Erroneously assuming that augmentation of systemic oxygen delivery would improve hemodynamic status, we decreased Qp/Qs < 1 when signs of low cardiac output became apparent. Indeed, samples with Qp/Qs ≤ 1 were coupled with worse hemodynamic status, reflected by lower urinary output, and higher lactate levels and doses of inotropes required, even if lower arteriovenous saturation difference and higher oxygen excess factor indicated higher systemic oxygen delivery in this group. Total cardiac output and mean common atrial pressure are increased in samples with Qp/Qs ≤ 1, highlighting a hyperdynamic state and resulting in volume overload of the single ventricle, which may have contributed to higher mortality within this group. Thus, mathematical models were partially validated in the clinical setting, since, as proposed, systemic oxygen delivery indices increase as Qp/Qs falls below 1. However, enhancement of systemic oxygen delivery did not translate into improved hemodynamic status and lower lactate levels. Serum lactate values have been shown to predict major adverse events and mortality after operation for complex congenital heart disease [11]. In this study, lactate levels were a very useful monitoring instrument to signify stable hemodynamic status and adequate myocardial function after stage I palliation. Low systemic output failure with balanced or imbalanced Qp/Qs was accompanied by escalating lactate levels, denoting tissue oxygen deficit and energy synthesis from anaerobic metabolism. With rising lactate levels impending circulatory failure is to be suspected and immediate intervention to improve Qp/Qs and/or cardiac function is mandatory. These findings resulted in a change of postoperative management and hemodynamic targets at our institute (Table 4). Most importantly, Qp/Qs is adjusted at a level of 1.5 rather than of 1 and adequate perfusion and end-organ function is monitored by hourly collected lactate levels. When increasing lactate levels were recognized, patients were managed as follows: Balanced circulations with low total cardiac output: SaO2 of 75–80% and SvO2 of 40–55% with corresponding Qp/Qs between 1 and 2: with rising lactate levels, inadequate total cardiac output and low systemic oxygen delivery is assumed. Since Qp/Qs is already in optimal range, measures to improve systemic oxygen delivery concentrate on elevation of hematocrit and on augmentation of inotropic support. If aerobic conditions are still not achieved, ECMO support should be instituted. Unbalanced circulations with high systemic oxygen delivery: SaO2 of 75–80% and SvO2 above 60% with corresponding Qp/Qs ≤ 1: with rising lactate levels, measures that increase systemic vascular resistance or decrease pulmonary vascular resistance should be initiated to adjust Qp/Qs between 1 and 2. Unbalanced circulations with pulmonary overcirculation: SaO2 of 75–80% and SvO2 below 40% with corresponding Qp/Qs ≥ 2: with rising lactate levels, measures that decrease systemic vascular resistance or increase pulmonary vascular resistance should be commenced to adjust Qp/Qs between 1 and 2. We conclude that manipulations of pulmonary to systemic blood flow ratio influence hemodynamic status after the Norwood procedure. Even if maximal systemic oxygen delivery was seen with Qp/Qs ≤ 1, better hemodynamic status, end-organ function, lower lactate levels and mortality were recognized, when Qp/Qs was adjusted between 1 and 2. Measures which decrease Qp/Qs below 1 in order to increase systemic oxygen delivery should not be recommended.