Major clinical impact of patent foramen ovale after HeartMate3 implantation: periprocedural diagnosis and its pitfalls

Abstract

We report a clinically significant right-to-left intracardiac shunt through a patent foramen ovale, diagnosed during investigations for hypoxemia and left ventricular dilation on the late postoperative period of a HeartMate3 implantation. We discuss diagnostic pitfalls and haemodynamic influences in this scenario, as well as the possibility of successful percutaneous treatment.

Background

As more patients are enlisted for heart transplant without concomitant increase in the number of organ donors, long-term left ventricular assist devices (LVADs) might become an alternative for patients with advanced heart failure worldwide. Currently, the HeartMate3 (Abbott), a magnetically levitated centrifugal continuous-flow pump, is one of the most commonly implanted long-term LVADs. Its inflow cannula is surgically placed on the left ventricular apex, through which blood is drawn from the heart and pumped into the ascending aorta through the outflow cannula, improving systemic perfusion and allowing patients’ ambulation and return to daily activities.

Given the haemodynamic changes induced by this and all LVAD types, exclusion of intracardiac shunts is mandatory before implantation, as their presence might complicate clinical course. Strict protocol for intracardiac shunt exclusion should be standardized, by means of saline contrast and provocative manoeuvres during echocardiographic study, while acknowledging possible diagnostic influencers at all stages of evaluation.

Case presentation

A 43-year-old male with advanced heart failure (HF) and reduced left ventricular ejection fraction (LVEF) of ischaemic aetiology was repeatedly hospitalised for acute HF requiring inotropes (INTERMACS 4), despite guideline-directed medical and device therapy. While being enlisted for heart transplant, his recurrent admissions triggered the decision for long-term LVAD implantation as bridge-to-transplant, and he was electively admitted for HeartMate3 (HM3) implantation.

Presurgical transthoracic echocardiogram (TTE) (figure 1) displayed severely reduced LVEF (<20%), moderate right ventricular (RV) dysfunction, no significant valvular disease and a left ventricular (LV) apical thrombus despite warfarin therapy, none of which contraindicating HM3 implantation. Intracardiac shunts were excluded using saline contrast plus the Valsalva manoeuvre.

Figure 1.

Presurgical transthoracic echocardiogram. (A) Parasternal long-axis view. (B) 4-chamber apical view, (C) 3-chamber apical view, all showing severely dilated left ventricle. (D) 4-chamber apical view after contrast injection, showing apical thrombus (arrow).

HM3 implantation was successful, preceded by apical thrombus removal. Intraoperative transoesophageal echocardiogram (TEE) showed no evidence of intracardiac shunts by colour Doppler analysis. Appropriate LV decompression was guided by LV dimensions, interventricular septum position and intermittent aortic valve opening, while taking into consideration pre-existing moderate RV dysfunction, and was achieved at 4600 rpm with a final pump flow of 3.7 L/min.

The early postoperative period was uneventful: absence of severe hypoxemia or other ventilatory requirements allowed patient’s extubation on day 2, inotropic support with milrinone was stopped on day 4 and HM3 pump speed was gradually increased over the first few days, allowing the patient to ambulate for short distances without need for supplementary oxygen. However, after day 7, severely reduced exercise tolerance and oxygen desaturation ensued (<80% on room air, only slightly improved by supplementary oxygen). Mean blood pressure was adequate (70 mm Hg), as was heart rate (75 bpm); no signs of pulmonary or systemic congestion were noted on physical examination.

Differential diagnosis

In order to exclude potential LVAD and non-LVAD causes for exercise intolerance (box 1), blood tests, chest X-ray, ECG and TTE were performed.

Box 1. Differential diagnosis of reduced exercise tolerance following HM3 implantation.

Respiratory

• Alveolar hypoventilation (asthma / COPD exacerbation)

• Intrapulmonary shunt (alveolar oedema, atelectasis, pneumonia, pulmonary arteriovenous malformations)

• Pleura (pleural effusion, pneumothorax)

• Chest wall (reduced expansion in relation to pain, diaphragmatic palsy)

Cardiac

• RV failure

• Severe tricuspid or mitral regurgitation

• Significant aortic regurgitation

• Arrythmias

• Pulmonary embolism

• Intracardiac shunts (PFO, ASD, VSD)

• Cardiac tamponade

LVAD^-related

• Suboptimal LVAD support

• Inflow cannula obstruction (thrombus, suction events)

• Pump thrombosis or mechanical malfunction

• Outflow cannula obstruction (thrombus, kinking, increased afterload)

Other

• Hypovolemia

• Anaemia (including haemolysis)

• SIRS

• Deconditioning

• Significant anaemia was excluded and inflammatory markers were only mildly elevated.

• Chest X-ray was innocent for pulmonary or pleural pathology (figure 2A). No diaphragmatic palsy was seen.

• ECG showed normal sinus rhythm (figure 2B ), and telemetry revision showed no paroxysmal arrythmias.

• The main relevant finding on TTE was a dilated LV, with interventricular septum shift to the right and a small RV, at first interpreted as insufficient LVAD support and inadequate LV decompression. However, despite daily pump speed increments of up to 5600 rpm producing 4.7 L/min of pump flow, neither symptoms, hypoxemia nor TTE findings improved.

• Despite increasing venous return to the RV owing to higher pump flow, there were no signs of worsening RV function (neither clinical, laboratorial or echocardiographic).

• Using the same data on RV size and function, and taking into consideration the use of full-dose anticoagulation for pump thrombosis prophylaxis since postoperative day 3, a haemodynamically significant pulmonary embolism was considered unlikely. More so, moderate kidney dysfunction postponed CT pulmonary angiogram.

• No substantial valvular dysfunction was present, namely aortic regurgitation, that, if significant, could have increased recirculation of the blood pumped into the aorta back to the LV, increasing LV dimensions and decreasing systemic perfusion.

• LVAD malfunction was improbable, given no recent device alarms. Evident inflow cannulae obstruction was absent on TTE, and pump thrombosis was unlikely as haemolysis parameters were normal. Outflow cannula was not visualised on TTE evaluation.

• Only mild pericardial effusion was noted.

Figure 2.

(A) Chest X-ray (posteroanterior projection) showing cardiomegaly and no pulmonary or pleural alterations. HeartMate3 pump and driveline in situ, as well as single-chamber implantable cardioverter-defibrillator. (B) ECG, sinus rhythm at 70 bpm. Baseline artefact owing to HeartMate3 interference.

Given the unexplained clinical picture, a TEE was performed, confirming previous TTE findings and excluding outflow cannula obstruction. Finally, interatrial septum examination showed a patent foramen ovale (PFO), with a continuous, significant right-to-left shunt on Doppler colour analysis (figure 3 and videos 1–3), explaining the clinical picture of de novo hypoxemia. The shunt had been previously missed on preoperative TTE and intraoperative TEE. Given poor acoustic window, serial postoperative TTE evaluations also failed to identify the right-to-left shunt.

Figure 3.

Postoperative transoesophageal echocardiogram, showing right-to-left shunt through a patent foramen ovale (PFO). (A) 2D colour Doppler (bicaval view), arrow shows right-to-left shunt. (B) 3D reconstruction (right atrial view), arrow denotes PFO’s tunnel-like defect.

Video 1.

Download video file ^{(523KB, mp4)}

DOI: 10.1136/bcr-2021-241585.video01

Video 2.

Download video file ^{(557.2KB, mp4)}

DOI: 10.1136/bcr-2021-241585.video02

Video 3.

Download video file ^{(559KB, mp4)}

DOI: 10.1136/bcr-2021-241585.video03

Treatment

Percutaneous PFO closure was performed with a CeraFlex (Lifetech) device (figure 4). Subsequently, the patient experienced clinical improvement, hypoxemia was no longer noted and adequate LV decompression was achieved at a lower pump speed (5000 rpm producing 4.0 L/min of flow). The RV dimensions and pre-existing moderate dysfunction remained unchanged.

Figure 4.

Patent foramen ovale percutaneous closure (transoesophageal echocardiogram images). (A) CeraFlex device in situ (midoesophageal short-axis view). (B) Agitated saline contrast injection showing minimal residual shunt (midoesophageal 4-chamber view).

Outcome and follow-up

The 1-month follow-up TEE with agitated saline contrast showed no residual interatrial shunt.

At 12-month follow-up, the patient had NYHA class I symptoms, no heart failure admissions ever since and was still enlisted for heart transplant.

Discussion

After LVAD implantation, main haemodynamic changes include LV decompression with left-sided filling pressure reduction, improved systemic output and increased venous return, leading to lower left-sided and higher right-sided cardiac pressures. Intracardiac shunts may become deleterious after these LVAD-induced haemodynamic changes, promoting right-to-left shunting, hypoxemia and paradoxical embolisation; as such, their detection and correction in LVAD patients is mandatory.¹

In this patient, pre-implantation TTE was negative for a PFO. However, as shunt direction depends on relative pressures on both sides of the interatrial septum, tremendously elevated left-sided filling pressure owing to severe LV dysfunction left the PFO undiagnosed, despite saline contrast and the Valsalva manoeuvre (which increased right atrial pressure transitorily, but insufficiently for shunt diagnosis). This is a common false negative in patients with advanced HF and makes PFO diagnosis reliable only after device activation.²

Intraoperatively, TEE before and after HM3 activation and after weaning from cardiopulmonary bypass was also unremarkable, yet evaluation was performed by colour Doppler analysis only. Agitated saline contrast injection is notably more sensitive to detect any degree of right-to-left shunting when compared with colour Doppler analysis,³ and both methods are recommended early after device activation.⁴ Furthermore, provocative tests such as transitory PEEP increase in mechanically ventilated patients can improve TEE sensitivity significantly by increasing right atrial pressure and right-to-left shunt.⁵ In addition, pump speed leaving the operating room was optimised using TEE guidance, yet a ramp study (consisting on stepwise increments in pump speed in order to optimise pump flow) was not performed. If done, especially at a higher pump speed, it could have also unveiled the PFO, through LV filling pressure reduction and shunt intensification. In conclusion, absence of a strict intraoperative TEE evaluation protocol delayed PFO diagnosis, as saline contrast, provocative ventilatory manoeuvres and possibly ramp study would have recognised the shunt in the first place and allowed for its correction during the HM3 implantation procedure.

We also hypothesise that a meaningful right-to-left shunt was not immediately present postoperatively, but became clinically significant only after adequate haemodynamic conditions developed. HM3 implantation allowed LV decompression and increased systemic output and venous return, leading to RV preload increase with additional dilatation and dysfunction. All these changes, despite facilitating right-to-left shunt, were probably balanced at first, owing to initially low pump speed and inotropic support with milrinone. The shunt’s initially insignificant clinical impact was illustrated by the patient’s early extubation and ambulation without supplementary oxygen. Later on, RV and right atrium overload were further worsened by gradual pump speed increments and inotrope weaning, a combination of haemodynamic phenomena that led to an increasingly significant right-to-left shunt and to clinical deterioration after a seemingly uncomplicated early postoperative period. PFO diagnosis has been described in other acute RV overload/dysfunction scenarios, such as RV infarction and acute pulmonary embolism, all of them leading to a sudden right atrial pressure increase, right-to-left intracardiac shunt and hypoxemia, which characteristically responds poorly to increased inspiratory oxygen fraction.⁶

Right-to-left shunt pathophysiology is similar among LVADs and most reported cases presented with early severe hypoxemia and failure to wean from the ventilator, allowing for PFO’s prompt correction; our patient became symptomatic only later in the postoperative period, possibly for the aforementioned explanation. A similar late diagnosis was reported in a patient on biventricular mechanical support that became hypoxemic after weaning from temporary RV support, having a PFO percutaneously closed on day 13 post-LVAD implantation.⁷ An even later and less common clinical picture of a post-LVAD patent PFO has been described, presenting with position-induced intermittent hypoxemia (platypnoea–orthodeoxia syndrome) and successfully corrected by percutaneous approach.⁸ These cases raise awareness for a ‘hidden’ PFO diagnosis even during long-term follow-up.

When diagnosed outside the surgical implantation setting, medical management strategies to counteract right-to-left shunting include ventilation at low intrathoracic pressures, pulmonary vasodilation and careful volume management, as well as reduction of LVAD output.⁹ When hypoxemia remains clinically significant despite these measures, surgical, or more often, percutaneous PFO closure, poses a definitive solution. Successful PFO percutaneous closure has been previously reported with other devices other than HM3.^9–11

This case highlights the need for a thorough intraoperative exclusion of intracardiac shunts during LVAD implantation, while keeping in mind that right-to-left shunt magnitude and symptoms can be variable, owing to different preload conditions, concomitant RV support (mechanical or pharmacological) and suboptimal LVAD support, among others. A PFO should be immediately considered in a hypoxemic patient with LV dilation in the later postoperative stage, even after its initial apparent exclusion. When diagnosed, percutaneous closure is a safe and effective solution for uncorrected post-LVAD PFO, obviating the need for surgery, especially in bridge-to-transplant patients.

Learning points.

• Intracardiac shunts should be systematically sought and corrected in left ventricular assist device (LVAD) candidates.

• Preoperative echocardiogram may not be able to completely exclude a patent foramen ovale (PFO).

• Intraoperatively, agitated saline contrast plus provocative manoeuvres are recommended for adequate shunt exclusion.

• During follow-up after LVAD implantation, hypoxemia and LV dilation should be identified as red flags for a PFO, irrespective of the time of presentation after surgery.