Discussion

The management of late-diagnosed transposition of the great arteries with intact ventricular septum remains a challenge in paediatric cardiology and surgery, particularly in regions with socioeconomic limitations and delayed neonatal healthcare access. In our series of 24 patients, the BDG + pulmonary artery banding strategy enabled successful two-stage arterial switch operation in 96% of cases with only 8.7% requiring temporary ECMO and no late mortality, demonstrating its safety and efficacy in achieving adequate left ventricular retraining even in severely deconditioned ventricles from resource-limited settings. Without treatment, mortality can reach 90% in the first year due to persistent hypoxaemia and multiorgan failure.^{Reference Kang, de Leval and Elliott5} While timely arterial switch operation yields excellent outcomes in specialised centres,^{Reference Wernovsky, Mayer and Jonas6,Reference Rakha, Batouty, ElDerie and Hussein22} , delayed diagnosis beyond 2–4 months leads to left ventricular deconditioning, rendering primary arterial switch operation high-risk with reported mortality of 20–30% and frequent ECMO requirement.^{Reference Cevik, Karaci and Polat9,Reference Tan, Datuk Ja’apar and Sallehuddin14} Two-stage protocols address this by retraining the left ventricle, and BDG + pulmonary artery banding offers a physiologically balanced approach by increasing afterload without excessive preload, thus minimising complications associated with shunt-based methods. Our findings from a socioeconomically disadvantaged cohort add substantial evidence supporting this strategy in high-risk, late-diagnosed patients. Although shorter interstage intervals following left ventricular retraining have been proposed to minimise annular mismatch and potentially reduce early neo-aortic regurgitation, the optimal timing of arterial switch operation after BDG + pulmonary artery banding remains undefined. In our practice, progression to arterial switch operation was guided by functional and structural markers of left ventricular readiness rather than time alone. Therefore, the relatively longer interstage duration observed in this study reflects a strategy of ensuring stable and complete ventricular adaptation rather than delayed intervention.

Preoperatively, our patients demonstrated severe left ventricular regression, reflected by a median left ventricular mass index of 24.5 g/m² and banana-shaped left ventricular geometry in 75% of cases, which is extremely low considering the safe value (35 g/m²) for primary arterial switch operation.^{Reference Lacour-Gayet, Piot and Zoghbi16} The first-stage operation (BDG + pulmonary artery banding) resulted in significant improvement in oxygen saturation (from 75% to 85.5%) while maintaining moderate pulmonary gradients (median 26 mmHg). Postoperative recovery was quick in the great majority of patients, with a median ICU stay of 4.5 days without extensive needs for mechanical ventilation or inotropic support. Complications were rare, albeit one patient (4.17%) died due to sepsis. These results align with the series by Cevik et al., ^{Reference Cevik, Karaci and Polat9} who reported no mortality and comparable ICU stays and infection rates in a similar cohort. Our findings support the utility of BDG + pulmonary artery banding as a safe option to facilitate a stable haemodynamic environment that achieves left ventricular adaptation.

Following a median retraining duration of 15 months, left ventricular mass index increased markedly to 78 g/m², and left ventricular geometry normalised in all patients. Angiographic data showed systemic-level left ventricular pressure (≥90 mmHg) and McGoon ratios ≥1.8, indicating adequate pulmonary arterial development. Subsequent arterial switch operation was completed successfully in 23 patients with no mortality. Two patients (8.7%) required temporary ECMO for transient left ventricular dysfunction, both of whom recovered. At a median follow-up of 2.5 years, all patients remained in NYHA class I, with mild neo-aortic regurgitation in 52.17% and mild neo-pulmonary regurgitation in 13.04%, which are frequencies that agree with long-term outcomes in prior studies.^{Reference Duan, Sun, Dong, Du and Yan7,Reference Cevik, Karaci and Polat9,Reference Watanabe, Mainwaring, Carrillo, Lui, Reddy and Hanley10,Reference Schulz, Kelm and Weixler12,Reference Tan, Datuk Ja’apar and Sallehuddin14,Reference Yıldız, Ozturk, Onan, Tanıdır, Ergul and Guzeltas15,Reference Daoud, Nuri, Miette and Pomè17} The absence of reoperation or late ventricular dysfunction is a critical finding that demonstrates the reliability of this gradual retraining method.

Comparative literature further supports our findings. Schulz et al. ^{Reference Schulz, Kelm and Weixler12} observed left ventricular mass increases from 21 to 38.5 g/m² after combined pressure–volume retraining in congenitally corrected transposition of the great arteries, while Duan et al. ^{Reference Duan, Sun, Dong, Du and Yan7} reported higher morbidity and longer ICU stays using shunt + pulmonary artery banding. Similarly, Watanabe et al. ^{Reference Watanabe, Mainwaring, Carrillo, Lui, Reddy and Hanley10} reported a 42% mortality rate in late primary arterial switch operation without retraining, whereas Yıldız et al. ^{Reference Yıldız, Ozturk, Onan, Tanıdır, Ergul and Guzeltas15} achieved 0% mortality but required ECMO in 25% of cases. These data collectively demonstrate that abrupt loading strategies are needlessly risky compared with the gradual adaptation achieved with BDG + pulmonary artery banding.

To further elaborate on the results of pulmonary artery banding combined with Blalock-Taussig shunt versus pulmonary artery banding with BDG in late-diagnosed transposition of the great arteries with intact ventricular septum, both methods enable effective left ventricular retraining, but pulmonary artery banding + Blalock-Taussig shunt is associated with shorter durations (mean 98 days) at the cost of higher early morbidity, including prolonged ICU stays (significantly higher than BDG), extended inotropic support, and increased risks of complications such as pulmonary overcirculation, infection, and thrombosis.^{Reference Cevik, Karaci and Polat9,Reference Schulz, Kelm and Weixler12} In contrast, pulmonary artery banding + BDG promotes gradual left ventricular adaptation over longer periods (mean 117 days), resulting in fewer postoperative complications, shorter hospital stays, and stable oxygenation without excessive preload, while achieving comparable left ventricular mass increases (from ∼25 g/m² to ∼68 g/m²) and successful arterial switch operation with low ECMO needs.^{Reference Cevik, Karaci and Polat9,Reference Schulz, Kelm and Weixler12} In a retrospective cohort of 47 patients, the BDG group demonstrated reduced overall morbidity compared to the Blalock-Taussig shunt, supporting its role as a safer alternative for high-risk, late-presenting cases (Reference Cevik, Karaci and Polat9).

The physiological rationale behind BDG + pulmonary artery banding is its ability to stimulate adaptive hypertrophy while maintaining balanced haemodynamics. Pulmonary artery banding increases left ventricular wall stress within tolerable limits, activating mechanosensitive signalling cascades such as integrin-linked kinases, titin kinase domains, and the calcineurin–NFAT and ERK1/2 pathways, leading to concentric hypertrophy and enhanced contractile capacity.^{Reference Schulz, Kelm and Weixler12,Reference Rakha, Batouty, ElDerie and Hussein22,Reference Jensen23} Meanwhile, BDG decompresses the right ventricle by diverting superior vena cava flow directly to the pulmonary arteries, reducing systemic venous return and preventing volume overload.^{Reference Cevik, Karaci and Polat9,Reference Yıldız, Ozturk, Onan, Tanıdır, Ergul and Guzeltas15} This avoids the excessive preload and turbulent flow characteristic of shunt-based methods, which can provoke pulmonary overcirculation and vascular remodelling.^{Reference Kang, de Leval and Elliott5,Reference Lacour-Gayet, Piot and Zoghbi16} Potential complications related to BDG, including superior vena cava (SVC) syndrome and SVC dilation or aneurysm formation due to competitive flow between the Glenn circulation and the antegrade LVOT–pulmonary artery pathway, have been reported in the literature.^{Reference Chen, Tulloh, Caputo, Stoica, Kia and Parry24,Reference Panduranga, Thomas, Al-Maskari and Al-Farqani25} In our cohort, careful modulation of pulmonary artery banding and maintenance of controlled antegrade flow likely minimised excessive SVC pressure, and no clinical signs of SVC syndrome or SVC aneurysm were observed during follow-up. Maintaining distal pulmonary pressures at 25–30 mmHg ensures gradual left ventricular retraining, resembling the natural transition from fetal circulation.^{Reference Wernovsky, Mayer and Jonas6,Reference Nakazawa, Oyama and Imai26} Controlled afterload prevents maladaptive signalling through p38-MAPK and JNK pathways, which are associated with apoptosis and fibrosis in overstressed myocardium.^{Reference Castaneda, Norwood, Jonas, Colon, Sanders and Lang27,Reference Gatzoulis, Walters, McLaughlin, Merchant, Webb and Liu28} The resulting left ventricular remodelling favours pro-survival Akt–mTOR signalling and angiogenesis, supported by VEGF upregulation,^{Reference Schwartz, Gauvreau, del Nido, Mayer and Colan29,Reference Chatterjee, Miller, Cribbs and Law30} thereby explaining the excellent postoperative left ventricular performance and minimal ECMO requirement observed in our cohort.

Additionally, the balanced haemodynamics produced by BDG + pulmonary artery banding help mitigate the consequences of chronic hypoxaemia. By maintaining oxygen saturations around 85%, the protocol minimises HIF-1α-mediated fibrosis and oxidative stress injury.^{Reference Gatzoulis, Walters, McLaughlin, Merchant, Webb and Liu28,Reference Losay, Touchot and Serraf31} The low incidence of postoperative ventricular dysfunction may thus reflect the preconditioning benefits of moderate hypoxaemia and controlled mechanical stress, which enhance ischaemia–reperfusion tolerance through induction of heat shock proteins and anti-apoptotic Bcl-2 pathways.^{Reference Chatterjee, Miller, Cribbs and Law30,Reference Lopez, Colan and Frommelt32} Mild neo-aortic regurgitation observed in our patients likely results from sinotubular junction mismatch, which is a well-established outcome of arterial switch operation, with higher frequencies reported in the literature.^{Reference Lang, Badano and Mor-Avi33,Reference Marino, Lipkin and Newburger34} Coronary anomalies identified in 13% of patients were also successfully managed, made possible in part by the low-pressure BDG environment that reduces ischaemic risk relative to high-flow shunt systems.^{Reference Jensen23,Reference Li and Ma35}

BDG + pulmonary artery banding represents a particularly suitable approach for low-resource or socioeconomically disadvantaged populations where delayed diagnosis is common. The technique leverages the inherent regenerative potential of the paediatric myocardium, which retains robust stem cell activity and proliferative capacity.^{Reference Mainwaring, Patrick and Arunamata13,Reference Nakazawa, Oyama and Imai26} Our experience suggests that late-diagnosed patients can achieve functional recovery if retraining is performed under controlled afterload conditions. Therefore, it may be feasible to suggest that current age-based exclusion criteria may be outdated, particularly since individualised assessment based on left ventricular compliance and geometry appears to be a superior approach. The absence of overtraining or diastolic dysfunction in our series indicates that the procedure provides a self-adjusting mechanism that maintains adaptive stress within physiologic limits.^{Reference Wernovsky, Mayer and Jonas6,Reference Cevik, Karaci and Polat9,Reference Yıldız, Ozturk, Onan, Tanıdır, Ergul and Guzeltas15,Reference Nakazawa, Oyama and Imai26} Rapid retraining methods are known to predispose patients to arrhythmia and failure.^{Reference Boutin, Jonas, Sanders, Wernovsky, Mone and Colan20,Reference Sivakumar, Francis, Krishnan and Shahani36} The low postoperative pulmonary gradients in our patients further demonstrate balanced circulation that does not cause hypoplasia or stenosis—which can develop in shunt-based interventions.^{Reference Cevik, Karaci and Polat9,Reference Bishnoi, Patel and Agrawal11} Temporary ECMO use in two cases likely reflected the impact of baseline hypoxaemia, which was reversible, especially since both patients ultimately reached normal ejection fraction values.^{Reference Schulz, Kelm and Weixler12,Reference Lacour-Gayet, Piot and Zoghbi16}

Despite these encouraging results, certain limitations must be considered. The sample size is small (n = 24) and reduces statistical power, but this reflects the rarity of late-diagnosed transposition of the great arteries with intact ventricular septum. The single-centre retrospective design and relatively small sample size limit the generalisability of our findings and preclude definitive conclusions. Follow-up duration (1–4 years) may not be sufficient to assess longer-term complications such as aortic root dilation or arrhythmia.^{Reference Lang, Badano and Mor-Avi33,Reference Marino, Lipkin and Newburger34} Furthermore, echocardiography may underestimate subtle fibrosis or vascular anomalies, while angiographic or cardiac MRI are superior in this regard. Socioeconomic factors, including malnutrition and delayed somatic growth, were not systematically assessed but may have influenced left ventricular hypertrophy.^{Reference Jensen23} The lack of molecular or histologic assessment limits our ability to directly confirm the likely pathophysiologic mechanisms described in this study^{Reference Kang, de Leval and Elliott5,Reference Lacour-Gayet, Piot and Zoghbi16} Regular follow-up during the prolonged retraining period may be challenging in resource-limited settings due to restricted healthcare access, potentially increasing loss to follow-up risk. Detailed quantitative measurements of atrial septal defect size were not consistently available due to the retrospective nature of the study and therefore could not be systematically analysed. Another limitation of this study is the absence of a direct comparative cohort using a modified Blalock-Taussig shunt with pulmonary artery banding, which prevents institutional outcome comparison with this alternative retraining strategy. Future multicentre prospective studies incorporating biomarkers such as NT-proBNP and microRNAs (e.g., miR-208a) may improve predictive accuracy and long-term monitoring.^{Reference Rakha, Batouty, ElDerie and Hussein22,Reference Lopez, Colan and Frommelt32}