Discussion
To our knowledge, this is the largest study comparing shunt size to ASD dimension measured with TEE, and the first study to compare the shunt size to ASD dimension in adult population. The shunt size measured with dye dilution method showed good correlation with the actual ASD size measured by TEE or TEE with balloon sizing. The shunt size measured both by gold standard invasive oximetry and by non-invasive dye dilution showed an excellent agreement. Also, the agreement seems to be excellent across all sizes of shunt Qp:Qs values. Furthermore, dye dilution method showed good correlation with CMR shunt size measurement.
Right ventricular (with or without atrial) enlargement is considered the most important indicator for ASD closure.2 This enlargement is thought to reflect a large enough shunt to strain the right side of the heart when no other obvious reason for the enlargement is present. Shunt size assessment is needed in borderline cases or in cases with elevated right-sided pressure. Shunt Qp:Qs ratio of 1.5 or larger is accepted to be significant enough to cause expansion of the right side, however, there is no strong evidence supporting this statement.27 There are no large data available to show which size of shunt can cause ventricular enlargement and which size of ASD’s should be closed. Studies only show that patients benefit from closure of ASD’s large enough to cause right ventricular enlargement.28–31
Closing smaller ASD’s has never been studied in an endpoint driven manner. A large retrospective Danish cohort showed that patients with ASD’s not fulfilling closure criteria have increased cardiac mortality and they would benefit from medical re-examination since the haemodynamic properties of an ASD can change with ageing.32 At this time, we do not know that which is the best method to determine patients in need for ASD closure and if we should close defects with small dimensions and shunt sizes smaller than 1.5.
The shunt size is determined by the ASD dimensions, pressure gradient between left and right atria, ventricular compliance and cardiac output. Our study demonstrates that the shunt size correlates well with the dimensions of the ASD shown before only in a paediatric population.27 The current study shows that ASD size explains more than 2/3′s of the shunt size. Having said this, our evaluations in above cardiac measurements are limited. We do not have neither actual pressure gradient between left and RA nor cardiac output values. However, tricuspid regurgitation peak gradient measured with TTE gives a good estimation of pulmonary pressure, and it was in most cases normal or only mildly elevated.
Non-invasive dye dilution method has many advantages compared with other shunt size measurement techniques. It provides accurate results agreeing with invasive oximetry. In our comparison to invasive oximetry the dye was injected into the IVC or the RA. We can assume that the dye acts similar way as the dye injected to the antecubital vein.21
Dye dilution can detect shunt fractions as small as 1.15.17 In detecting patent foramen ovale the sensitivity was reported to be 76% and specificity 100%.20 Interobserver agreement of dye dilution detection in small shunts has previously shown to be excellent. The method is also time-efficient and cost-efficient. One measurement takes only about 15 min, and the price per study is relatively low, being about one third of the price of the CMR in our hospitals price list.
The dye dilution method has also limitations. Fragmented dye boluses can cause erroneous results. Significant valve regurgitations may influence to the Qp:Qs value. In our material there were no significant valve regurgitations. In atrial fibrillation the variance of heart rate and stroke volume impacts the shape of the dye dilution curve. Thus, more measurements are needed to interpolate the mean value. Dye dilution gives a good estimate of shunt size but does not yield information regarding the location of the shunt in the cardiovascular system.
Invasive oximetry has been the standard method for assessing the Qp:Qs ratio in intracardiac shunts. The invasive nature of the study is considered its biggest downside. Other limitations are lack of sensitivity, requirement of a cardiovascular steady state during sample collection, and calculation of MVO2: normal variability of blood oxygen saturation in the right heart chambers is strongly influenced by the magnitude of Qs.5 24 The cardiac output spontaneously varies at rest, which introduces inaccuracies, especially if there are delays between blood samplings at various sites.25
CMR is suggested as a good alternative for shunt size evaluation. It is available in the most hospitals; however, capacity and costs are the usual limiting factors to use it. Studies demonstrate good correlation in shunt size assessment between CMR and invasive oximetry, however, based on rather small populations (n=10–50).3 6 33–36 Also, there are some limitations in using MRI for shunt size evaluation: interpretation of the shunt is difficult with irregular rhythms, pulmonary and aortic pathology disturb the calculations, Qp/Qs ratio is overestimated if the assessment of the systemic flow is done distal to the coronary ostia in the ascending aorta,35 and breathing can influence the values obtained by CMR.17 Regardless of the above limitations, CMR is widely used in shunt size evaluation.
In this study, no correlation between radionuclide angiography and ASD size was found. The radionuclide evaluations were performed in small medical centres, with a limited amount of evaluations annually. This may cause difficulties in executing and interpreting the radionuclide measurement and may have influenced on our results.
This study has potential limitations. It was done in a retrospective manner and the study was not designed to compare dye dilution to CMR or radionuclide method, therefore, it is hard to make conclusions of superiority of one method over the others. Second, we employed two populations to show correlation both to the invasive oximetry and to the ASD size measured with TEE. We acknowledge that haemodynamic parameters would be important in the study, however, these parameters are not available in either of the populations. In 12 months control visit four patients had a significant residual shunt, however, they did not undergo additional tests to quantify residual shunt seen in dye dilution.