Dyspnoea and hypoxaemia after lung surgery: the role of interatrial right-to-left shunt

Statistics

Whenever appropriate, the two-tailed paired or unpaired t-test, Levene's test for equality of variance and Fisher's exact test were used to compare haemodynamic parameters, such as mean right atrial pressure, mean left atrial pressure and mean pulmonary arterial pressure, within each group, as well as their differences between groups. Data are reported as mean±sd. A p-value of <0.05 was considered significant throughout.

Type of lung surgery

The surgical procedures involved in group A (table 1⇑) were: right pneumonectomy in 24 patients (67%), left pneumonectomy in eight (22%) and right lobectomy in three (8%). In one patient, who underwent thoracotomy due to lung cancer, the type of operation was not reported. The types of surgical procedure in group B (table 2⇑) were: right pneumonectomy in six patients (60%), right lobectomy in three (30%) and left pneumonectomy in one (10%).

Overall, the right side of the chest was more frequently (36 out of 45 (80%)) involved than the left (nine out of 45 (20%)), and lung cancer was the disease affecting all patients.

Symptomless period

Table 1⇑ shows the cases in which dyspnoea occurred ≥1 month following lung surgery; their symptomless period ranged 1–31 months (mean 6±7 months).

Table 3⇑ shows the cases in which dyspnoea occurred within 1 month after lung surgery; their symptomless period ranged 1–7 days (mean 3±3 days).

Prevalence of congenital abnormality

Regarding the prevalence of congenital abnormality of the right atrium causing right-to-left shunt, PFO was more frequent by far than ASD: 78% (28 out of 36) of patients in group A, and 100% in group B. Regarding the ASD abnormalities, all listed in table 1⇑ (eight out of 36 (22%) patients), the interatrial communication was of the secundum type in five patients, whereas the type was not specified in two patients. In one patient, an atrial septal aneurysm was found at the level of the fossa ovalis.

Type of investigation for the diagnosis

Regarding diagnostic work-up in these patients, 16 out of the 36 (44%) patients in group A underwent the 100% oxygen breathing test. In a few of them, the test was performed in both the recumbent and the upright position. In group B, four out of the 10 (40%) patients underwent the 100% oxygen breathing test.

A perfusion lung scan, which is usually performed to rule out pulmonary embolism in such patients, was obtained in 17 out of 36 (47%) patients in group A and five out of 10 (50%) patients in group B. Nine out of the 17 (56%) perfusion lung scans performed in patients from group A showed no perfusion abnormality, whereas, in the seven other patients, the lung scan showed an extrapulmonary distribution of the radioactive tracer in the brain, thyroid, kidneys and spleen, suggesting the presence of right-to-left shunt 40. Interestingly, in one of these seven patients, the pattern of extrapulmonary distribution of radioactive tracer was neglected, and, in another patient, the extrapulmonary uptake of the radiotracer became evident 1 month after the initial evaluation, when deterioration of respiratory symptoms had occurred. Of the five lung scans performed in the 10 patients from group B, three did not show any perfusion abnormality, and two showed extrapulmonary uptake of the radioactive tracer.

Tests for the diagnosis of right-to-left intracardiac shunt

Among the 36 patients in group A, intracardiac shunt through PFO or ASD was demonstrated by inferior and/or superior vena cava angiography in 17 (47%) patients, contrast transthoracic or transoesophageal echocardiography in 15 (41%) and right heart catheterisation in two (6%), and, in one (3%) patient each, by magnetic resonance imaging and open heart surgery.

Regarding the 10 patients in group B, the final diagnosis of intracardiac right-to-left shunt was made by contrast transthoracic or transoesophageal echocardiography in eight (80%) patients and right heart catheterisation in two (20%).

Overall, the techniques more frequently used for the diagnosis of intracardiac right-to-left shunting in this series of 46 patients were, in order, angiography (17 patients), contrast transoesophageal echocardiography (12) and contrast transthoracic echocardiography (11).

Posture-dependent abnormalities in gas exchange

As shown in table 1⇑, platypnoea–orthodeoxia syndrome was present in 33 out of the 36 (92%) patients in group A, and it occurred ≥1 month after lung surgery. In 19 of these subjects, the finding of platypnoea–orthodeoxia was supported by sitting–supine changes in P_a,O2 or S_a,O2 (table 2⇑) according to the definition reported above 39. Of the 10 patients who complained of acute dyspnoea shortly after lung surgery, three manifested platypnoea–orthodeoxia syndrome and sitting–supine changes in P_a,O2 or S_a,O2 (table 3 and 4⇑).

Haemodynamic measurement

The information regarding haemodynamic parameters is fragmentary. From the available data shown in table 2⇑ (group A), it appears that mean right atrial pressure was 4.3±3.4 mmHg in 27 out of the 36 (75%) patients, mean left atrial pressure was 6.0±3.5 mmHg in 22 (61%) and mean pulmonary arterial pressure was 14.3±3.7 mmHg in 26 (72%; table 5⇓). Thus, in the patients in group A, the mean pressure gradient between the left and right atrium was 1.07±3.3 mmHg, and the mean pulmonary arterial pressure was normal (table 5⇓).

In group B (table 5⇑), mean right atrial pressure was 7.6±3.7 mmHg in eight out of 10 (80%) patients, whereas mean left atrial pressure was 7.1±3.6 mmHg in seven (70%) patients and mean pulmonary arterial pressure 21.8±10.7 mmHg in six (60%). Interestingly, the mean right atrial pressure in patients in group B was significantly higher than that of patients in group A (7.6±3.7 versus 4.3±3.4 mmHg, respectively; p<0.05; table 5⇑). The mean pulmonary arterial pressure in patients in group B (21.8±10.7 mmHg) was also higher than that of patients in group A (14.3±3.7 mmHg), although the difference did not attain significance (p = 0.15).

Closure of interatrial communication

Overall, the methods used to close a PFO or ASD in these 46 patients were: open-chest surgical closure in 32 (70%) patients, transcatheter permanent implantation of a device placed across the interatrial communication 23 in eight (17%) and a pulmonary arterial catheter with a balloon inflated in the left atrium for temporary closure in one (2%; table 2⇑). In the five (11%) other patients, four in group A (table 2⇑) and one in group B (table 4⇑), the interatrial communication was not closed. Indeed, patient No. 4 showed marked clinical improvement once diuretics and digoxin were withdrawn 10, patient No. 11 was followed up for 8 months after diagnosis and the surgical intervention was still under consideration at the time of the report 15, patient No. 14 was deemed too ill to undergo surgery 5 and patient No. 30 did not undergo surgery owing to a very poor prognosis due to the progression of lung cancer 25. Patient No. 5 in group B did not undergo surgical intervention, in part due to the spontaneous partial improvement of clinical symptoms, but mainly due to a poor prognosis 34.

Follow-up studies

In table 2⇑, the follow-up of patients of group A is reported. Overall, four out of the 36 (11%) patients died, two of them a few days after surgery.

Sixteen (50%) patients were followed-up for a mean period of 26±25 months (range 1.5–96 months). In eight (25%) patients, the follow-up ended at the time of their discharge from the hospital, in four (12.5%) it ranged 2–10 months, and, in the remaining four (12.5%), the follow-up period was not reported (table 2⇑).

Regarding patients in group B (table 4⇑), death occurred in five out of 10 (50%) patients, with two dying 1 day and 3 months after transcatheter PFO closure, respectively. Of the five patients who survived, three were followed-up until hospital discharge, one was followed-up for 6 months, one for 21 months, after PFO closure (table 4⇑).

Mechanisms involved in right-to-left shunt

The type and site of pulmonary resection do not appear to affect the time of onset of right-to-left shunt. Indeed, it occurred after right pneumonectomy, as well as after resection of a few pulmonary segments.

Similarly, the nature of the congenital abnormality, be it PFO or ASD, does not in itself explain the time of occurrence of right-to-left shunting. Indeed, both of these congenital abnormalities were usually asymptomatic before pulmonary resection 18.

By grouping patients as a function of the symptomless interval, it was found that platypnoea–orthodeoxia syndrome was present in 92% of patients of group A (table 1⇑), in whom the symptomless interval averaged 6±7 months. On the contrary, in the 10 patients of group B (table 3⇑), in whom the symptomless interval was 3±3 days, platypnoea–orthodeoxia syndrome was present in three out of the nine (33%) patients for whom these data were available (table 3⇑). Thus, the syndrome was significantly more frequent in patients in whom respiratory symptoms developed several months after lung surgery (p<0.001; Fisher's exact test).

The late occurrence of symptoms in patients with platypnoea–orthodeoxia (group A; table 1⇑) may be the result of relatively slow anatomical remodelling after lung resection. Indeed, the obliteration of the post-pneumonectomy space usually takes 3 weeks to 7 months 41. The shift of the mediastinum towards the operated side may be extensive when the obliteration of the pleural space is complete. In this case, the interatrial stretching can be conspicuous 42 and right-to-left shunting through a PFO or an ASD may develop (mainly in the upright position) despite normal right atrial pressure 8. The interatrial stretching seems to be the mechanism of the streaming of blood from the inferior vena cava to the left atrium in the absence of a pressure gradient, particularly in the presence of mediastinal distortion, when the right atrium is shifted away, while the inferior vena cava remains fixed in position 16. This mechanism appears operative in patients of group A, despite the fact that the mean left atrial pressure was higher than the mean right atrial pressure (table 5⇑). However, it has been suggested recently that the mobility of the fossa ovalis membrane might favour and enhance the flow towards the foramen ovale 43. This additional mechanism seems to be operative during mid-systole, when a pronounced motion of the atrial septum toward the left atrium occurs 44. Furthermore, as shown in table 5⇑, the mean pulmonary arterial pressure was normal (14.3±3.7 mmHg). Therefore, the occurrence of right-to-left shunting through a pressure gradient mechanism appears unlikely in these patients.

In group B, the mean right atrial pressure was higher than the mean left atrial pressure in four of the seven patients in whom these data were available (table 4⇑). By comparing the mean right atrial pressure between the two groups (table 5⇑), it appears that the mean right atrial pressure in most patients of group B was significantly higher than that found in patients of group A (7.6±3.7 versus 4.3±3.4 mmHg; p<0.05). Moreover, the mean pulmonary arterial pressure was higher in patients of group B than in those of group A.

Thus, in this case, the shunt could occur as a consequence of a sudden increase in pulmonary vascular resistance and a decrease in right ventricular compliance 45 following surgical reduction of the pulmonary vascular bed 18. In addition, the intra- and perioperative fluid overload, which is known to impair right heart function in pulmonary resection 46, 47, could also act through the same mechanism in these patients.

Interestingly, three of the 10 patients in group B showed platypnoea–orthodeoxia syndrome, although it has been demonstrated that this complication more frequently occurs as a late post-pneumonectomy complication.

In case No. 4 (group B), who underwent left pneumonectomy, the mean right atrial and pulmonary arterial pressures were higher than the corresponding mean pressures of patients in group B (tables 4⇑ and 5⇑). Therefore, the sudden increase in flow in the right lung may have caused an interatrial right-to-left shunt in this patient, with a further increase in the fraction of shunted blood in the sitting position. In case No. 6 (group B), the mean pressure in both atria was identical and the right-to-left shunt could occur on moving to the sitting position due to a gravity-dependent increase in blood flow 48. Thus, the pressure gradient mechanism seems to be the trigger of this clinical feature, even in two of the three cases who showed posture dependency of shunt shortly after surgery. Conversely, six out of the seven (86%) patients with available data in group B exhibited a baseline mean right atrial pressure higher than, or at least equal to, that of the left atrium.

Diagnostic and therapeutic management

Once the suspicion of right-to-left shunt is raised, the 100% oxygen breathing test may be useful, particularly when it is performed in the sitting and supine position. Indeed, it appears that 20 out of the 46 (43%) patients reported here had P_a,O2 while breathing 100% oxygen that were compatible with the presence of right-to-left shunting.

A perfusion lung scan may also show findings suggestive of right-to-left shunt, such as extrapulmonary distribution of radioactive tracer in the brain, thyroid, kidneys and spleen, particularly when the radiotracer injection is performed with the patient in the sitting position 25, 29.

In order to confirm the diagnosis of interatrial right-to-left shunt, right heart angiography was the technique more frequently used until 1988 28. Nowadays, it has been progressively replaced by contrast transthoracic or transoesophageal echocardiography, performed in 29 (63%) patients in the present series. Indeed, the latter technique is far less expensive and less invasive than angiography, although equally reliable.

Regarding the procedure used to close the interatrial shunt, open-chest surgery was the technique of choice in 32 out of the 46 (70%) patients of the present series. Since the mid-1990s, open heart surgery has been progressively replaced by transcatheter device implantation across the interatrial communication. This therapeutic option has been successfully applied in 78 patients from the French PFO registry who had platypnoea–orthodeoxia syndrome of various aetiology 49.

Follow-up of patients

Follow-up data show that nine patients died, four (11%) from group A and five (50%) from group B. Of these patients, only four died within a few days after interatrial communication closure, three of them after surgery and one after transcatheter device implantation.

The high rate of death of patients in group B is probably accounted for by their poor clinical conditions, as suggested, on the basis of the available data, by their very low P_a,O2 and/or S_a,O2. The occurrence of death in this group of patients appears independent of the method used to close the interatrial communication, be it surgery or transcatheter device implantation.