Mechanical support of pulmonary blood flow as a strategy to support the Norwood circulation-lumped parameter model study

Abstract

Open in new tabDownload slide

OBJECTIVES

We hypothesize that mechanical assistance of the pulmonary blood flow in a Norwood circulation can increase systemic blood flow and oxygen delivery. The aim of the study was to compare haemodynamics of an unassisted Norwood Blalock–Taussig shunt circulation with a mechanically assisted pulmonary flow-based Norwood circulation, using a lumped parameter computational model.

METHODS

A neonatal circulatory lumped parameter model was developed to simulate a Norwood circulation with a 3.5-mm Blalock–Taussig shunt in a 3.5-kg neonate. A roller pump circulatory assist device with an inflow bladder was incorporated into the Norwood circulation to mechanically support the pulmonary circulation. Computer simulations were used to compare the haemodynamics of the assisted and unassisted circulations. Assisted and unassisted models with normal (56%) and reduced ejection fraction (30%) were compared.

RESULTS

Compared to the unassisted Norwood circulation, the systemic flow in the assisted Norwood increased by 25% (ejection fraction = 56%) and 41% (ejection fraction = 30%). The central venous pressure decreased by up to 3 mmHg (both ejection fraction = 56% and ejection fraction = 30%) at a maximum pulmonary assist flow of 800 ml/min. Initiation of assisted pulmonary flow increased the arterial oxygen saturation by up to 15% and mixed venous saturation by up to 20%.

CONCLUSIONS

This study demonstrates that an assisted pulmonary flow-based Norwood circulation has higher systemic flow and oxygen delivery compared to a standard Norwood Blalock–Taussig shunt circulation.

INTRODUCTION

Norwood operation is the most utilized surgical stage 1 palliation strategy for patients with hypoplastic left heart syndrome (HLHS). Over the last 3 decades, various surgical modifications to the procedure have been described, but the basic principles of the operation have not changed. Although the surgical mortality has dramatically reduced since its introduction in the early 1980s, the survival has plateaued over the last 10 years [1]. Despite the advances made in operative and perioperative care, the Norwood stage 1 operation continues to be associated with a significant risk of mortality and morbidity. The procedure currently is associated with an operative mortality of about 15% [1], a 20% rate of serious neurological complications and a 5-year survival of 65% [2]. Strategies to protect and support myocardial function in the immediate postoperative period could potentially improve the surgical outcomes. With the Norwood operation being the standard of care for treating HLHS, it would be extremely challenging to clinically implement alternative procedures and treatment strategies. Strong laboratory data in form of computational fluid dynamic studies, acute and survival animal studies supporting alternate strategies would be required before clinical trials can be planned. As a first step to investigate a novel shunted single-ventricle mechanical support strategy we propose to use a lumped parameter computational model to test feasibility and generate preliminary data.

We hypothesize that mechanical assistance of the pulmonary blood flow (Fig. 1A) in a Norwood Blalock–Taussig shunt (NBTS) circulation can augment systemic flow (⁠$Qs$⁠) and increase oxygen delivery. This strategy proposes using a mechanical circulatory support (MCS) device to create the missing ventricle in the shunted single-ventricle circulation. The aim of the study was to compare haemodynamics of an unassisted standard NBTS (normal and reduced ejection fraction) with a mechanically assisted pulmonary blood flow (⁠$aQp$⁠)-based NBTS (normal and reduced ejection fraction) using a lumped parameter computational model (LPM).

MATERIALS AND METHODS

Ethical statement

Institutional Review Board approval was not required as clinical patient data was not utilized.

Haemodynamics and gas exchange model

Our previously well validated neonatal circulatory LPM [3, 4] was modified to analyse haemodynamics of the novel $aQp$-based NBTS (Fig. 1B). The target physiological parameters that represent a typical Norwood stage 1 patient were obtained based on published clinical data.

The constructed network consisted of major peripheral organs modelled as vascular elements, with compliant chambers and linear resistances connecting them. Mathematical modelling [5] was used to simulate the blood flow through the multi-compartmental LPM.

In addition to the pulsatile haemodynamic parameters, oxygen exchange between the compliance chambers of the cardiovascular system and the CO₂ transport were simulated by using a gas transport model [4].

Norwood stage 1 shunt haemodynamics

Standard Blalock-Taussig (BT) shunt characteristics were implemented using a validated [6] patient-specific model. A 3.5-mm diameter BT shunt was used to simulate the flow (⁠$Qt$⁠) supplying the pulmonary circulation through the shunt. Shunt resistance was determined through the Hagen-Poiseuille equation. In the LPM, hypoplastic left ventricle to underdeveloped aorta connection (aortic valve) was deactivated and a neo aorta was modelled that is driven by the single (right) ventricle. Ductus arteriosus was removed from the base model and the left atrial chamber was deactivated to obtain a common atrium.

Roller pump-based Q_p assist device

With the aim to use a simple and clinically relevant pump that can provide temporary postoperative circulatory support in a neonate a roller pump was utilized for the simulation. To couple a standard cardiopulmonary bypass roller-pump to the post-op NBTS, we utilized our previous approach developed for simulating Fontan circulatory assist [7]. Pump inlet was connected to the right atrium (RA) and outflow was directed to pulmonary artery (PA) chamber.

To prevent the negative inlet pressure or collapse of RA that may occur due to pump activity, a safety inflow bladder (25 cc) was incorporated into the pump inlet [8]. Safety bladder was mathematically modelled by increasing the RA compliance (decreased the elastance) to restrict its collapse due to proximal pump suction.

Single-ventricle model

To simulate postoperative ventricular dysfunction, reduced EF was simulated. $Emax$ and $Emin$ are reciprocals of the single ventricle’s systolic and diastolic compliances, respectively. Therefore, systolic and diastolic compliances of the ventricle were changed to reduce the stroke volume and restrict the ventricular capacity through its elastance function.

Simulated assisted post-operative conditions

Body weight of a full-term neonate was assumed as 3.5 kg (body surface area = 0.234 m²). Normal EF was chosen as 56% for the base model and it was reduced to 30% to simulate postoperative ventricular dysfunction.

After unassisted case was validated through clinical data (Results Section 1), MCS device was implanted to assist pulmonary flow (Q_p) in the NBTS and its effect on critical haemodynamic parameters: systolic/diastolic systemic blood pressure, central venous pressure (CVP), pulmonary artery pressure (PAP), systemic and pulmonary flow (Q_s and Q_p, respectively), arterial and mixed venous oxygen saturation levels (⁠$SaO2$ and $SvO2$⁠, respectively) were studied.

All post-operative haemodynamic parameters for the $aQp$-based NBTS were simulated with varied pump flow rates; $Qpump$ of 200 ml/min (case 1), 400 ml/min (case 2), 600 ml/min (case 3) and 800 ml/min (case 4).

Statistical analysis

A computational patient cohort of NBTS was generated using in silico trials and digital-twin technology [11–13]. Continuous random variables were chosen to be heart rate, systemic vascular resistance, pulmonary vascular resistance (PVR), systemic artery compliance (⁠$Csa$⁠), PA compliance (⁠$Cpa$⁠) and pump flow rate (⁠$Qpump$⁠), based on our earlier sensitivity studies and normally distributed in cohort. These variables are normally distributed in our digital cohort based on standard deviations available in literature [14–16] (Table 1). A cohort of 2000 in silico neonates with NBTS was created under the assumption of normality. Of these, 50% were categorized as having normal EF (56%) and other 50% having low EF (30%).

We compared the LPM outputs: (i) pulmonary flow, (ii) systemic flow, (iii) shunt flow, (iv) PAP, (v) Aortic pressure, (vi) CVP, (vii) EF and (viii) Q_p/Q_s by simulating continuous variables of randomly selected patients. Data comparison was executed separately between 2 categories of results as unassisted and assisted early post-Norwood periods for normal and low EF conditions.

For the comparison n = 35 patients were randomly picked from both normal and low EF categories of the digital cohort. Two group pre-test/post-test approach was used as the paired t-test (setting $α$ = 0.05) to compare the unassisted and assisted NBTS.

To assess the correlation between the unassisted and assisted haemodynamics, r-values were calculated through the Pearson method due to the normality of the chosen continuous random variables.

RESULTS

Model validation

Simulated postoperative haemodynamics were compared with the reported clinical data on postoperative haemodynamics of an NBTS.

LPM simulated the systolic/diastolic (mean) systemic blood pressure as 69/42 (56) mmHg, which is similar to the reported systemic blood pressure of 70/40 (55) mmHg [17]. Cardiac index, the total blood delivered to the systemic and pulmonary sides, was simulated as 3.9 l/min/m², which is in the clinical results range (4.2 ± 1.2 l/min/m²) obtained early after stage 1 palliation [18]. In both normal and low EF conditions, pulmonary-systemic flow ratio (⁠$Qp/Qs$⁠) was calculated as 1.41, which corresponds well with the reported [19, 20] $Qp/Qs$ of 1.4–1.7.

PAP and CVP were simulated as 17.5 mmHg and 7.5 mmHg, respectively. Early postoperative mean PAP measured in 39 patients with Norwood circulation was between 12 and 16 mmHg [18]. Simulated CVP also has a good correlation with reported clinical data of 6–12 mmHg [18–20]. $SaO2$ and $SvO2$ were simulated as 81% and 65% and were comparable to the recorded in vivo oxygen saturations [21] of 77% and 58%, respectively.

Simulated dynamic ventricular, aortic and pulmonary pressure and flow patterns including shunt flow are presented in Fig. 2, with the obtained pressure–volume (PV) loops for normal and low EF conditions. A good correlation is obtained between the simulated and published pressure-flow waveforms [14] and PV loops [22] of the single-ventricle patients.

A sensitivity analysis was also conducted to observe the effect of the systemic vascular resistance/PVR, systemic venous compliance (⁠$Csv$⁠) and single-ventricle elastance on haemodynamics with and without ventricular assist device (VAD) operation. We observed that the ventricular elastance function is the most significant parameter on haemodynamics, while $Csv$ is the least effective one except considering the CVP. In addition, there is no significant instability on haemodynamics observed with the VAD usage.

Systemic blood flow (Q_s) and shunt flow

In Fig. 3, the effect of $aQp$-based Norwood circulation on Q_s, Q_t, Q_p and $Qp/Qs$ has been illustrated for the generated computational cohort. Based on pulsatile simulations, initiation of $aQp$ decreases the $Qt$ and increases the Q_s and Q_p for both EF of 56% and 30%. For a simulated maximum $Qpump$ of 800 ml/min (case 4), the $Qt$ decreases by up to 120 ml/min and the common atrial pressure decreases by 3 mmHg.

Although the net $Qp/Qs$ increases with the $aQp$⁠, the workload on the single ventricle remains the same. In the assisted circulation, the decrease in Q_t is transferred to the systemic circulation leading to an increase in the Q_s. Considering both effects together, the systemic flow can be increased by 25% (EF-56%) and 41% (EF-30%) for a standard Norwood neonate under the maximum $Qpump$ operation.

Figure 3 demonstrates that the effect of $aQp$ on Q_s and Q_t is similar for both normal EF (56%) and reduced EF (30%). It also shows that to gain 30 ml/min from the Q_t, $Qpump$ should be about 200 ml/min. Since there is a considerable pressure gradient between the aortic arch and PA compliance chambers, the shunt flow is reduced but not eliminated even at maximum $aQp$⁠.

Ventricular pressure–volume curves

PV loops of the assisted (⁠$Qpump$ = 800 ml/min) and unassisted NBTS are shown in Fig. 4. In the assisted circulation, the single-ventricle generates higher aortic pressure at a lower or same stroke volume. $aQp$ increases the mean aortic pressure by 5 mmHg in low and 8.5 mmHg in normal EF conditions. In addition, Fig. 4 demonstrates that both end-diastolic and end-systolic volumes increase with the pump assist. With the preload conditions being identical, $aQp$-based circulation can generate a higher stroke volume compared to the unassisted circulation.

Pulmonary artery and central venous pressures

The PAP and CVP changes under varying $aQp$ flows were simulated for normal and low EF patient cohorts (Fig. 5). In simulated unassisted case, CVP increased from 7 to 10 mmHg when the EF was reduced to 30%. At maximum $aQp$ of 800 ml/min, the venous filling pressure decreased by 3 mmHg and the PAP increased by 15 mmHg. Since the MCS device used to provide $aQp$ was operated at the same conditions for both normal and reduced EF conditions, the decrease in CVP and the increase in PAP observed did not depend on the EF of the patient. However, the benefit achieved with $aQp$ in terms of decreasing fill pressure and increasing $Qs$ was higher in patients with low EF.

Risk of common atrial collapse and venous suck down

In the normal and reduced EF models collapse of common atrium was observed at roller pump flows higher than 1370 and 2050 ml/min, respectively.

Arterial and mixed venous oxygen saturation levels

Figure 6 summarizes the effect of $aQp$ on SaO₂ and SvO₂ for normal and low EF patient cohorts. With $aQp$⁠, the $SaO2$ increased by 15% and $SvO2$ increased by 20%. With $aQp$ close to physiologically normal, $SaO2$ and $SvO2$ were achievable.

Statistical evaluation of the proposed support strategy

Statistical analysis (Table 2) demonstrated significantly (P < 0.0001) better haemodynamics with assisted NBTS (low and normal EF) compared to unassisted NBTS (low and normal EF).

DISCUSSION

The single-ventricle parallel circulation in a patient with HLHS is inherently unstable and can rapidly decompensate during periods of increased physiological demand. Norwood operation subjects the patient to cardiopulmonary bypass and cardioplegic arrest but does not correct the underlying shunted single-ventricle physiology. Postoperatively the recovering single ventricle continues to experience significant volume overload and hypoxia. The trauma of surgery significantly increases the postoperative physiological demand of the patient. The recovering neonatal ventricular myocardium is most vulnerable to stretch and distention injury in the initial 12–24 h post-surgery. A combination of all these factors increases the risk of postoperative low cardiac output state and cardiac arrest. Following Norwood operation, the need for veno-arterial extracorporeal membrane oxygenation (VA-ECMO) support and extracorporeal-cardiopulmonary resuscitation is 10–20% [23].

Myocardial failure and low cardiac output state are the most common causes of death in the postoperative period [24]. Of the patients who survive the operation about 10–15% develop systemic ventricular dysfunction and heart failure by 6 years of age [25]. Strategies to protect and enhance myocardial performance in the postoperative period could potentially improve outcomes.

The use of stem cell therapy to recover and regenerate myocardial function following the stage 2 palliation has been described but may be of little benefit immediately after stage 1 operation. Elective MCS therapy to assist the myocardial function in the immediate postoperative period could improve outcomes. VA-ECMO support following the Norwood operation has been shown to be associated with an increased risk of mortality [26]. Routine use of the ECMO circuit without the oxygenator as a systemic VAD in the post-Norwood setting has been described [27]. This strategy still exposes the patient to the large surface area of the ECMO circuit and the risk of systemic thromboembolism. Conventional systemic MCS strategies subject the heart to low preload and flow-induced increased afterload. The effect of these loading conditions on the neonatal single-ventricle myocardium is unknown. Several animal studies have demonstrated VA-ECMO flow induced left ventricular dysfunction in structurally normal hearts [28]. Various mechanical support options including elimination of the Norwood stage 1 circulation and creation of a mechanically assisted primary bidirectional cavopulmonary shunt circulation [29] has been proposed as a strategy to improve outcomes in patients with HLHS. The lack of a reliable mechanical assist device to support a cavopulmonary circulation and the need to cannulate the neonatal superior vena cava limits the clinical application of this strategy [29]. A radical shift in the management strategy of single-ventricle heart defects is required to improve survival in this complex cohort of patients.

With the proposed $aQp$ support strategy, the $Qp$ is almost entirely supported by the MCS device decreasing the workload of the systemic ventricle. Computational analysis demonstrated that $aQp$ reduces the systemic to PA shunt flow unloading the systemic ventricle. With $aQp$ the cardiac output generated by the systemic ventricle is preferentially directed into the systemic circulation. Therefore, a higher $Qs$ is generated at the same or lower venous filling pressures. The $aQp$-based NBTS has lower common atrial pressure with higher $Qs$ and systemic blood pressure when compared with the unassisted NBTS. Increased CVP is not well tolerated in a neonatal circulation and even a small decrease in the venous pressure can significantly improve end-organ perfusion. Although the net $Qp/Qs$ ratio increases, the $Qp$ supported by the native systemic ventricle decreases and the assist device prevents the ventricular distension and pulmonary oedema by decompressing the common atrium. This paradox also helps in increasing the systemic saturation and systemic oxygen delivery that should help in postoperative myocardial and end organ recovery. The increase in $Qs$ was higher in low EF model compared to the normal EF model demonstrating higher benefit with $aQp$ support in patients with ventricular dysfunction.

The presence of competitive flow has traditionally been considered a risk factor for graft/shunt thrombosis. It is important to note that even at max $Qpump$ of 800 ml/min, the $Qt$ is reduced but not eliminated reducing the risk of shunt occlusion during $aQp$ support. Using an electrocardiogram gated pulsatile pump to provide aQ_p flow during diastole could potentially help in mitigating the risk of shunt thrombosis and needs further study. More objective information on the effect of $aQp$ on the $Qt$ can be obtained from mock loop simulations and animal model studies. Although exclusion or clipping of the BT shunt during $aQp$ support would completely volume unload the systemic ventricle and eliminate the problem of competitive flow, it would make the circulation $aQp$ dependent. To provide redundancy in the circulation in case of mechanical pump failure, we believe a patent BT shunt is essential during $aQp$ support. Until stage 2 palliation is achieved, the systemic ventricle continues to support both the $Qp$+$Qs$ and it is currently not possible to reduce the workload of the systemic ventricle during the inter-stage period. With an $aQp$-based shunted single-ventricle circulation, a two-ventricle-based circulation (systemic ventricle + VAD) can be achieved to support postoperative recovery. VAD specifically designed to support $aQp$-based single-ventricle circulation is currently not available but several promising devices underdevelopment like the BioVAD [30] and Torvad [31] could potentially be used to provide both temporary and durable $aQp$ support.

Continuous flow centrifugal pumps tend to overflow and cause venous suck down when the afterload is low and may not be ideal for $aQp$ support. High $aQp$ flow may could lead to venous collapse, depletion of preload and low $Qs$⁠. A roller pump or a pulsatile flow pump would be required to provide controlled $aQp$ into the pulmonary circulation. To provide predictable and controlled $Qp$ support we used a standard roller pump with 25 cc safety bladder incorporated into its inflow to prevent venous collapse. The $aQp$ flow will have to be titrated to an optimal common atrial pressure that would provide ideal loading conditions for the systemic ventricle without excessively lowering the preload. Flows up to 800 ml/min have been simulated in this study but lower $aQp$ flows may be sufficient to provide adequate $Qp$ and augment the Q_s. Although it is not uncommon to a have $Qp:Qs$ of up to 2:1 after a Norwood operation [19] providing excessive $Qp$ can damage the pulmonary vascular and increase the PVR. In the simulation, the roller pump was able to successfully support the $Qp$ without causing venous suck down at the simulated flows.

Limitations

Although computational models are increasingly being used to study and model congenital heart defects, accurate simulation of a complex single-ventricle circulation can be extremely challenging. Advocating for clinical application of this strategy just based on computational model studies would be hard to justify. This study however does provide basic preliminary data to warrant further research to explore this concept. The findings of this study can be further validated in mock loop circulation studies and finally in a shunted single-ventricle animal model [32]. The current study demonstrates the utility of $aQp$ support in an NBTS. Implementation of this strategy in a Norwood Sano circulation would require the use of a valved Sano shunt. The feasibility of creating an $aQp$-based Norwood-valved Sano circulation needs to be assessed in an animal model. The proposed aQ_p support strategy only provides temporary support to optimize and support the haemodynamics during the vulnerable immediate postoperative period. After withdrawal of $aQp$ support, the systemic ventricle will again be subjected to significant volume overload until stage 2 palliation is completed. Durable MCS devices to support the patient during the entire interstage period are currently not available. The digital patient cohort is based on realistic clinical variances available from literature but needs to be verified with real patient data.

Successful validation of the proposed $aQp$-based shunted single-ventricle circulation could lead to a new strategy to mechanically support a shunted single-ventricle circulation. Compared to VA-ECMO, $aQp$ exposes the patient to a much smaller surface area of the extracorporeal circuit without need for oxygenator potentially reducing the inflammatory response, anticoagulation needs and transfusion requirement. Directing the mechanical assist device outflow into the pulmonary circulation does not increase the systemic afterload and should also significantly reduce the embolic burden that is delivered to the systemic circulation potentially reducing the embolic stroke risk. This strategy could potentially be used as bridge to postoperative recovery or as salvage therapy to rescue patients who are unable to wean from cardiopulmonary bypass or VA-ECMO support post Norwood operation. Future computational fluid dynamic studies using patient specific 3-D anatomy and animal studies in a neonatal pig animal model would be required to further validate this strategy and plan clinical translation of this approach.

CONCLUSION

This in silico study demonstrates that an $aQp$-based NBTS has higher $Qs$ and oxygen delivery compared to a standard NBTS.

Conflict of interest: none declared.

Data Availability Statement

The study data available on request.

Author contributions

Syed Murfad Peer: Conceptualization; Investigation; Methodology; Project administration; Supervision; Visualization; Writing—original draft; Writing—review & editing. Canberk Yildirim: Data curation; Investigation; Software; Validation; Writing—review & editing. Manan Desai: Data curation; Methodology; Validation; Writing—review & editing. Karthik Ramakrishnan: Writing—review & editing. Pranava Sinha: Conceptualization; Methodology; Writing—review & editing. Richard Jonas: Conceptualization; Supervision; Writing—review & editing. Can Yerebakan: Resources; Validation; Writing—review & editing. Kerem Pekkan: Conceptualization; Investigation; Methodology; Supervision; Validation; Writing—review & editing.

Reviewer information

European Journal of Cardio-Thoracic Surgery thanks Katarzyna Januszewska, Gaetano D. Gargiulo, Hitendu Hasmukhlal Dave and the other, anonymous reviewer(s) for their contribution to the peer review process of this article.

Presented at the Congenital Heart Surgeons’ Society 2021 Meeting, Chicago, USA. October 24, 2021.

REFERENCES

ABBREVIATIONS

ABBREVIATIONS
 
• CVP

  Central venous pressure

 
• EF

  Ejection fraction

 
• LPM

  Lumped parameter computational model

 
• MCS

  Mechanical circulatory support

 
• NBTS

  Norwood Blalock–Taussig shunt

 
• PA

  Pulmonary artery

 
• PAP

  Pulmonary artery pressure

 
• PV

  Pressure–volume

 
• PVR

  Pulmonary vascular resistance

 
• RA

  Right atrium

 
• VAD

  Ventricular assist device

 
• VA-ECMO

  Veno-arterial extracorporeal membrane oxygenation