Article Text

Download PDFPDF
Tetralogy of Fallot: from fetus to adult
  1. Elliot A Shinebourne1,
  2. Sonya V Babu-Narayan2,
  3. Julene S Carvalho3
  1. 1Department Paediatric Cardiology, Royal Brompton Hospital, London, UK
  2. 2Adult Congenital Heart Unit, Royal Brompton Hospital, London, UK
  3. 3Brompton Fetal Cardiology, Royal Brompton Hospital, London, UK
  1. Correspondence to:
    Dr Elliot A Shinebourne
    Paediatric Cardiology, Royal Brompton Hospital, Sydney Street, London SW3 6NP, UK; e.shinebourne{at}

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Tetralogy of Fallot (ToF) occurs in approximately 1 in 3600 live births and accounts for 3.5% of infants born with congenital heart disease. Surgical repair was first introduced in the 1950s and there is now a large population of adults with repaired tetralogy. Many of the short term aspects of management have been resolved, although whether symptomatic neonates should undergo primary repair or first be palliated remains debatable. In 2001, independently validated data pooled from all 13 centres performing cardiac surgery in the United Kingdom indicated a 97% survival one year after operation.1 Other reports indicate that of patients alive 30 days after operation there is a 98% 20 year survival, and of those operated on as children 30 year survival is above 90%.2

This has focused attention on long-term follow up as many patients with repaired ToF are now middle-aged. Given that this population includes patients well beyond adolescence they have also outgrown the term “grown-up congenital heart disease” (GUCH) which could be considered patronising. They have adult congenital heart disease and will need specialised care into old age.

The diagnosis of ToF can now be made antenatally with particular benefit for the fetus with severe right ventricular outflow tract obstruction (RVOTO). Postnatally the need for a prostaglandin infusion to maintain ductal patency can be anticipated and hence potential hypoxic cerebral damage can be avoided.


Anterior craniocaudal or cephalad deviation of the insertion of the muscular outlet septum together with hypertrophy of trabeculations on the infundibular free wall constitute the essential features of ToF. Instead of nestling between the anterior and posterior limbs of the trabecula septomarginalis, the outlet septum is displaced and typically fuses with the anterior limb. This brings the aorta over the ventricular septum so that it has a biventricular origin and accounts for the malalignment ventricular septal defect (VSD). The displaced outlet septum, together with the trabeculations of the parietal or free wall, produce muscular sub-valvar RVOTO. The effective pulmonary valve ring is usually smaller than the aortic valve orifice and the valve itself may or may not be stenosed.

From the surgeon’s viewpoint3 the roof of the VSD is formed by the attachments of the overriding aortic valve leaflets with the muscular ventriculo-infundibular fold. The anterior margin is formed by the outlet (infundibular) septum fusing with the anterior limb of the trabecula septomarginalis, the remainder of which forms the floor of the defect. In approximately 80% of cases the postero-inferior aspect is formed by fibrous continuity between aortic, mitral and tricuspid valves and is similar to the perimembranous VSD found in the absence of aortic override. The penetrating bundle of His perforates the central fibrous body through the region of aortic tricuspid continuity, where it may be overlaid by remnants of the membranous septum. This membranous flap can be used safely to anchor stitches, but deeper stitches could damage the bundle of His producing heart block.

In approximately 20% of cases the postero-inferior margin is muscular, formed by fusion of the posterior limb of the trabecula septomarginalis with the ventriculo-infundibular fold. This muscle bundle then separates the conduction tissue from the rim of the VSD. In Japan a third type of defect, namely doubly-committed subarterial, is more frequent than in the West. There are problems of definition as the outlet septum is absent, but as the hearts are otherwise similar they are usually included in the ToF group. In addition, other VSDs may rarely be present—for example, muscular inlet defects and particularly with Down’s syndrome a complete atrioventricular septal defect. Other variants include absent pulmonary valve syndrome and pulmonary atresia with multiple aorto-pulmonary collaterals (MAPCAs).


The diagnosis of ToF cannot be suspected prenatally unless views of the outflow tracts are obtained—being overlooked if screening is based solely on the four-chamber view. Leftward rotation of the cardiac axis, however, is not infrequent and may distort the four-chamber view, prompting referral for specialist scan. A significant number of referrals, however, are due to abnormal views of the great vessels, particularly at the level of the upper mediastinum (three-vessel view).4 Fetal diagnosis is based on the presence of a large VSD with aortic overriding, a smaller diameter of the pulmonary outflow tract and a relatively larger aorta. While the site of infundibular obstruction due to antero-cephalad deviation of the outlet septum can be imaged prenatally (fig 1), Doppler velocities in the pulmonary artery (PA) are in general normal or only mildly increased in contrast to post-natal findings.5 In the mid second trimester fetus, peak instantaneous velocities of ∼1–1.2 m/s across the RVOT are consistent with the diagnosis. Contrary to children, a hypertrophied right ventricle is not a prenatal feature, as in fetal life the right and left ventricles have similar afterloads and, therefore, similar wall thickness.

Figure 1

 Short axis view at the level of the aortic valve obtained from a 21-week fetus with tetralogy of Fallot (ToF). Note the large perimembranous ventricular septal defect (*), displaced outlet septum and narrow right ventricular outflow tract. ant, anterior; Ao, aorta; OS, outlet septum; PA, pulmonary artery; post, posterior; RA, right atrium, RV, right ventricle,

In general, the size of the main PA and its ratio to the ascending aorta in the mid trimester fetus with ToF reflects the severity of outflow tract obstruction (fig 2 and 3), being significantly smaller in those who will present as a duct-dependent pulmonary circulation. In contrast, the branch PAs tend to be of normal or near normal size in mid gestation. Growth of the pulmonary vessels throughout pregnancy may be normal or reduced, the latter being more frequently associated with more severe forms of outflow tract obstruction.5 The arterial duct is smaller than normal in fetuses with ToF and forward (right to left) flow is usually maintained. The presence of left to right ductal flow indicates severe outflow tract obstruction or pulmonary atresia.

Figure 2

 Upper mediastinal views obtained from a fetus at 22 weeks with ToF of good anatomy. Note the continuity between the right ventricle (RV) and pulmonary artery (PA) and the presence of confluent, good size branch. AAo, ascending aorta; DAo, descending aorta; LPA, left pulmonary artery; RPA, right pulmonary artery.

Figure 3

 Upper mediastinal views obtained from a fetus at 28 weeks with ToF and associated pulmonary atresia. Note the presence of small, confluent branch pulmonary arteries (left panel) and reversal of flow through the arterial duct (right panel). Ao, aorta; D, duct; DAo, descending aorta; LPA, left pulmonary artery; RPA, right pulmonary artery.

Thus serial scans to document fetal pulmonary artery growth and ductal flow patterns are essential for counselling and appropriate management of the newborn baby.

Absent pulmonary valve syndrome accounts for 3–6% of all cases of ToF. Vestigial or absent valve leaflets lead to pulmonary regurgitation as well as stenosis. Characteristically, the PAs are dilated and the arterial duct is usually but not always absent (figs 4 and 5). The presence of fetal hydrops and ventilator dependence in the neonate seem to correlate negatively with survival whereas the size of the PAs does not. The enlarged PAs, which can be seen from the early second trimester of pregnancy (fig 4), compress the bronchial tree and account for respiratory symptoms postnatally. The cause of hydrops and poor outcome in some fetuses may be related to right ventricular dysfunction, perhaps secondary to important pulmonary regurgitation in fetal life. Initial data on biventricular function using the Tei index in affected fetuses shows significantly higher values in non-survivors suggesting that global ventricular dysfunction plays a central role in outcome. In those cases where the arterial duct is patent, the branch PAs are not importantly dilated (fig 5).

Figure 4

 Transverse views of the fetal chest obtained from two fetuses at 22 weeks with ToF and absent pulmonary valve syndrome. Note absence of the arterial duct (left panel), grossly dilated branch pulmonary arteries, increased velocity across the right ventricular outflow tract, and pulmonary regurgitation. Ao, aorta; LPA, left pulmonary artery; RPA, right pulmonary artery; PA pulmonary artery; RV right ventricle.

Figure 5

 Transverse views of the fetal chest obtained from two fetuses at 22 weeks with ToF and absent pulmonary valve syndrome. Note the patent arterial duct, mildly dilated main pulmonary artery and relatively normal size branch. Ao, aorta; D, duct; LPA, left pulmonary artery; PA, pulmonary artery; RPA, right pulmonary artery.

Upon fetal diagnosis of ToF, associated extracardiac abnormalities need to be excluded. Fetal karyotyping to include analysis of microdeletion of chromosome 22q11.2 is offered to all families, even in the presence of an apparently isolated cardiac diagnosis.


A small proportion of patients with ToF or its variants, particularly those with a right aortic arch, absent pulmonary valve syndrome or pulmonary atresia and major aortopulmonary collateral arteries, may have some degree of immune deficiency (low T cells) as part of di George or cardio-velo-facial syndrome. This is an autosomal dominant condition caused by a deletion in the long arm of chromosome 22 (22q11.2 deletion) and is found in approximately 15% of ToF patients. They may, however, be otherwise phenotypically normal. Transfusion associated graft versus host disease (TA-GVHD) is defined as an attack by transfused immunocompetent T cells, which the host is unable to reject, on host HLA antigens producing pancytopenia and secondary sepsis. Though usually fatal it can be prevented by transfusing only γ-irradiated blood. There is probably only a significant risk if the CD 4 T cell count is < 1000 cells/mm3 or CD 8 cells < 800/mm3. Patients with ToF should be screened for 22q11.1 deletion, although this is unreliable in defining immune status as some with the deletion will have normal immune function while others with apparent di George syndrome may be negative for the deletion but have low T cells. It seems logical therefore to send blood for T cell subsets before surgery, and use irradiated blood in the group with low CD4 and/or CD8 numbers. The disadvantage of irradiated blood is the risk of hyperkalaemia if stored over 24 hours and for this reason in our own unit,6 unlike others, we use irradiated blood only when T cells are low.


The severity of RVOTO dominates clinical presentation. When severe at birth, there is a duct-dependent pulmonary circulation and a prostaglandin E infusion may be required to maintain ductal patency. Moderate RVOTO gives rise to a systolic murmur in an asymptomatic child often detectable on day 1, in contrast to neonates with an isolated VSD where the murmur typically appears later when the pulmonary vascular resistance falls. Cyanosis may develop between 6–8 months as infundibular stenosis increases, producing a right to left shunt. Presentation may also be with a hypercyanotic attack from infundibular shut-down. Diagnosis can be confirmed by cross-sectional echocardiography which demonstrates not only the ventricular septal defect, aortic override and RVOTO but also coronary anatomy (3% have a coronary anomaly), the presence of systemic pulmonary collaterals if present, the anatomy of the proximal PAs, and other associated anomalies.


Surgical strategies have evolved. Repair is via a transatrial, transpulmonary approach where possible. Transannular patches to relieve RVOTO are used sparingly if possible to preserve pulmonary valve function. There is no benefit in delaying elective repair of ToF after the end of the first year. Currently, symptomatic infants over six months also undergo primary rather than staged repair. What about younger infants? Some recommend primary repair no matter how young the patient. Others recommend primary repair down to the age at which the individual hospital risk of a shunt and subsequent repair is less than that of primary repair. Twenty years ago, that was 9 months when a transannular patch was needed and 6 months when it was not. More recently, the Toronto group changed their approach from one of initial palliation in the symptomatic infant to one of elective primary repair around the age of 6 months or earlier if clinically indicated.7 There were no deaths since the change of protocol although the intensive care stay was longer in those operated under 3 months of age. They concluded that optimal age for elective repair was 3–11 months. As the new protocol was non-contemporaneous, additional factors (for example, use of ultra-filtration) may have had a major influence, but nonetheless a case for earlier primary repair was made.

Hirsh et al8 reported on primary repair in 61 consecutive neonates. This paper presents excellent surgical results with 95% five year actuarial survival, yet raises concerns longer term. In this group of patients with the intracardiac anatomy of ToF, 24 had pulmonary atresia and six non-confluent pulmonary arteries. Twelve required a right ventricle to pulmonary artery conduit, 49 a transannular patch and 52 were operated on using hypothermic circulatory arrest. The five-year freedom from reoperation was 58%. While no neurological complications were detected, later neurodevelopmental impairment has been described after any form of cardiopulmonary bypass, but particularly in neonates in whom profound hypothermia and circulatory arrest was used. Even in the absence of overt convulsions electroencephalographic seizures have been shown to occur in 5–20% of neonates undergoing biventricular repair on cardiopulmonary bypass, the occurrence of which is a long-term predictor of adverse long-term neurodevelopmental sequelae. Because of the anatomy, many of these children will require further cardiopulmonary bypass procedures to replace a conduit. Also, a transannular patch leads inevitably to pulmonary regurgitation, the long-term harmful effects of which are detailed later. A protocol of palliative shunt and later repair is likely to reduce the number of bypass procedures, although whether this will reduce the frequency of transannular patching and hence severity of pulmonary regurgitation is uncertain.

Another factor in decision-making relates to the technical skill of the surgeon in performing a modified Blalock-Taussig shunt. Jahangiri et al9 have shown that a properly constructed shunt causes growth of both ipsilateral and contralateral PAs, rarely distorts the PA, and facilitates subsequent repair. Not all centres have such favourable results. In the 1970s and 1980s PA size was considered a major determinant as to whether primary repair of tetralogy could be carried out. This remains the case for older children but is probably not so important for the neonate or young infant where small size of PAs may simply reflect low flow, the PAs having the capacity to dilate with increased flow. We continue to endorse a similar policy to Fraser et al10 whereby symptomatic neonates under 4 kg are shunted with primary repair performed at 6–12 months in the remainder. The final arbiter, however, for which protocol is optimal will only come with long-term follow-up of infants operated on in the recent era.


Another factor that can influence recovery from surgery and duration of intensive care, as well as the late response of the right ventricle to pulmonary regurgitation, is the development of what has been termed restrictive right ventricle physiology. While postoperative recovery following repair is usually rapid with patients leaving the intensive care unit within 24–48 hours, some develop a low cardiac output and require considerable circulatory support. Doppler studies in these patients show anterograde diastolic flow in the PAs coinciding with atrial systole both during inspiration and expiration (fig 6). As the pulmonary vascular resistance is low, atrial contraction causes flow to be transmitted through a poorly compliant or “stiff” right ventricle into the PAs in such a way that if pulmonary incompetence is present, its duration is shortened. Not only does atrial contraction cause antegrade flow in the PAs during diastole, but retrograde flow can be detected in the superior vena cava, again, throughout the respiratory cycle. What causes this phenomenon is not yet fully established. It is more common in patients in whom a transannular patch has been inserted but is not directly related to age at operation. Recently, in adults with repaired ToF late gadolinium cardiovascular magnetic resonance (CMR) was used to study myocardial fibrosis (fig 6). Restrictive right ventricle physiology was found to be associated with more extensive late gadolinium enhancement. It is probable (perhaps as expected) that increased fibrosis or scarring, whether occurring due to an early insult or late progression, is the underlying substrate for restrictive right ventricle physiology.11

Figure 6

 Images obtained from a 44-year-old patient undergoing echocardiography and cardiac magnetic resonance imaging on the same day. (A) Doppler trace showing significant pulmonary regurgitation and antegrade flow in the pulmonary artery in late diastole (arrowed), indicative of restrictive physiology. (B) Right ventricular outflow tract (RVOT) view showing extensive late enhancement of an akinetic region of the RVOT (short arrows) and enhancement of the trabeculated myocardium (long arrow). (C) Left ventricular outflow tract view showing late enhancement in the RVOT free wall (short arrows) and along the moderator band (long arrow). (D) Short axis view showing extensive endomyocardial late gadolinium enhancement (short arrows).


The risk of pregnancy in repaired patients depends on haemodynamic status but is low, approaching that of the general population, in patients with good underlying haemodynamics and no symptoms. In patients with significant residual RVOTO, severe pulmonary regurgitation and right ventricular dysfunction, the increased volume load of pregnancy may lead to heart failure and arrhythmias. Any left ventricular dysfunction present in turn increases the likelihood of complications during pregnancy and requires separate consideration.

All patients with ToF should be offered specialist cardiological counselling pre-conception, genetic counselling and screening according to patient preference, fetal echocardiography and follow-up during pregnancy.

Burn et al12 studied patients undergoing cardiac surgery in 12 major centres in the UK before 1970. They identified 395 patients with ToF who had 223 children, seven of whom (3.1%, around three times that of the normal population) had congenital heart disease, not necessarily tetralogy. The risk of recurrence was greater for mothers than fathers. Clinically (not by genetic testing), none of these patients had di George syndrome, a condition with which the recurrence risk would be much higher due to the autosomal dominant pattern of inheritance.


Pulmonary regurgitation is common following repair of ToF. Although well tolerated for some years, formal exercise testing shows a negative correlation between pulmonary regurgitant fraction measured angiographically and exercise ability.13 In the longer term it leads to progressive RV dysfunction, ventricular tachycardia and sudden cardiac death.14 CMR is now the optimal method for non-invasive quantification of pulmonary regurgitation and assessment of ventricular function. Clinical assessment of pulmonary regurgitation alone is unreliable as a short diastolic murmur may be found both in very mild or extremely severe regurgitation. Pulsed wave Doppler echocardiography has been used to calculate regurgitant fraction as has the duration of retrograde flow as a percentage of total diastolic time using continuous wave Doppler. This technique has good agreement with CMR, but the latter remains the optimal technique for full assessment of the right ventricle.

Through-plane phase velocity mapping can be used to quantify pulmonary regurgitant fraction (fig 7). When pulmonary valve function is absent, a regurgitant jet may not be seen as most of the flow is laminar. In the setting of “free” pulmonary regurgitation and no valve function the pulmonary regurgitant fraction is typically not more than 40%.15 The functional consequences of pulmonary regurgitation differ from that of severe aortic regurgitation for two reasons.16 Firstly, as in the Fontan circulation, forward pulmonary flow can occur in the absence of a right ventricle due to the effects of inspiration and also because the left ventricle ejects blood out of the thoracic cavity, creating a negative pressure that sucks blood into the PA. Secondly, the pulmonary microvascular resistance is low and situated relatively near to the heart. As a result right ventricular systole moves blood through the pulmonary microcirculation, including capillaries, into the low pressure pulmonary veins. Flow that passes through alveolar capillaries does not pass back again in diastole, there being no significant reversal of gradient. Thus, the magnitude of the regurgitant fraction is limited unless there are additional factors such as branch PA stenosis which may increase regurgitation leading to progressive right ventricular dysfunction. In this context, balloon angioplasty and stenting may delay the need for pulmonary valve replacement. Other factors may also contribute to pulmonary regurgitation including the pulmonary valve annulus size, the compliance of the pulmonary arterial tree and reconstructed RVOT, and the function of the pulmonary microvasculature.15

Figure 7

 (A) RVOT cardiovascular magnetic resonance cine image obtained from a patient with repaired ToF with significant late pulmonary regurgitation. The red dotted line illustrates the through-plane in which a non-breath-hold phase encoded velocity map was acquired. (B) Flow curve obtained from the same patient. Through integrating areas containing forward and reverse flow, a pulmonary regurgitation fraction of 34% was calculated.

As CMR is independent of geometrical assumptions it has advantages over cross-sectional echocardiography for assessment of right ventricular mass, volumes and function. It also allows evaluation of RVOT akinetic areas. In addition to velocity mapping pulmonary regurgitant fraction can be measured by comparison of right and left ventricular stroke volumes. Serial measurements of RV volume are useful in decision-making, but there are problems of reproducibility as well as debate as to how to measure the akinetic area frequently seen in the RVOT.

The robust nature and reproducibility of volume measurements are well documented for the normal heart and for the abnormal left ventricle, but not for the abnormal right ventricle as in repaired ToF. The boundaries of the right ventricle are the pulmonary and tricuspid valves, but these can be difficult to define when the pulmonary valve is no longer functioning or visible. Additionally, how should we measure or analyse the akinetic area commonly seen in the RVOT following repair of ToF? Large RVOT akinetic areas may over-contribute to end-systolic volume measurements and in so doing mask smaller, progressive changes in the body of the right ventricle, which are potentially of more importance. Certainly they are a factor impacting on right ventricular volume that is unrelated to pulmonary regurgitation itself. Suggestion has been made that these areas and their corresponding volumes be excluded from volumetric measurements. However, akinetic outflow tract regions do not delineate a separate independently functioning compartment of the right ventricle. They negatively impact upon systolic function and may be a focus of arrhythmia. In our centre we include the outflow region in the right ventricular volume and measure the maximum linear extent of the akinetic area which is then taken into account in decision-making.


If there were such a thing as a perfect pulmonary valve prosthesis, the decision regarding timing of pulmonary valve replacement would be easier. Though there appears to be little difference between the longevity of a xenograft, homograft or stented bioprosthesis, all of these have a finite lifespan. This puts the younger patient on a potential pathway of serial reoperation, especially as younger age is a risk factor for earlier redo pulmonary valve replacement and the longevity of a second homograft may be shorter than the first. Thus there may still be a place for a mechanical valve in selected patients. Balancing the risk of late right ventricular dysfunction, arrhythmia and sudden cardiac death in these patients against the finite lifespan of a bioprosthetic valve and risks of surgery, leads to debate over the optimal timing of pulmonary valve replacement. There is an increasingly held view that we are operating too late and in so doing risk irreversible damage to the right ventricular myocardium. Conversely operating too early may consign patients to multiple surgical interventions.

Pulmonary valve replacement is a low risk procedure with a perioperative mortality of 1–4% and 10 year actuarial survival of 86–95%. The risk is increased with impaired right ventricular systolic function. In a mixed group of adults and children freedom from further valve replacement was 81% at five years, 58% at 10 years, and 41% at 15 years. Currently, timing of pulmonary valve replacement is most controversial in the asymptomatic patient with regurgitation greater than 30% by CMR. Indications are summarised in table 1 which includes patient preference. Paediatric cardiologists follow many children with mild residual RVOTO and moderate pulmonary regurgitation without thought of intervention. In contrast the adult perspective is that pulmonary valve replacement should be carried out before symptoms develop and RV function deteriorates.

Table 1

 Repaired tetralogy of Fallot and significant pulmonary regurgitation*: indications for pulmonary valve replacement

Following pulmonary valve replacement, within six months there is a rapid decrease in right ventricular end-diastolic volume,17 maintained a year later. Systolic function may also improve.17 Recently Bonhoeffer’s group introduced transcatheter pulmonary valve implantation, using a stent-mounted bovine jugular venous valve.18 Most patients had mixed stenosis and incompetence following previous surgical implantation of a valved conduit. This technique is not yet applicable to severe pulmonary regurgitation and a dilated RVOT, but further developments will widen the group of patients for whom this technique can be used.


Sudden death occurs in about 6% of patients over the long term2 or 4.5 per 100 patients years, but can it be predicted and modified? Impaired ventricular function, a residual VSD, residual RVOTO or severe pulmonary regurgitation are all associated with a poor prognosis possibly due to arrhythmias. The risk of sudden death in the asymptomatic patient is not predicted by the presence of non-sustained ventricular tachycardia (VT) or other ventricular arrhythmia on 24 hour ECG monitoring, nor is there apparent benefit in long term antiarrhythmic treatment. QRS prolongation of 180 ms or more, however, has been shown to be related to right ventricular dilatation, sustained ventricular arrhythmias and sudden death.19 Following reduction in right ventricular volume after pulmonary valve replacement, a reduction in QRS duration has been observed which may lessen the risk of arrhythmic complications. Clearly addressing target haemodynamic lesions remains a mainstay of arrhythmia risk reduction, but other aspects need consideration.

In a multicentre study20 programmed ventricular stimulation has been shown to be of diagnostic and prognostic value with both inducible monomorphic and polymorphic VT predicting future symptomatic VT and sudden cardiac death. This group, as well as patients with symptomatic VT or syncope, may warrant an implantable direct current defibrillator. In the future, assessing degree of myocardial fibrosis11 may contribute to risk stratification.

In conclusion, as many patients with repaired tetralogy of Fallot enter middle or old age, the need for continued surveillance and informed cardiac care is self-apparent.

Additional references appear on the Heart website—


Dr Sonya V Babu-Narayan was supported by the British Heart Foundation


View Abstract


  • *Also Cardiovascular Magnetic Resonance Unit, Royal Brompton Hospital, London, UK;

  • †Also Fetal Medicine Unit, St. George’s Hospital, London, UK

  • In compliance with EBAC/EACCME guidelines, all authors participating in Education in Heart have disclosed potential conflicts of interest that might cause a bias in the article

Linked Articles

  • Miscellanea
    BMJ Publishing Group Ltd and British Cardiovascular Society