Gas Exchange Detection of Exercise-Induced Right-to-Left Shunt in Patients With Primary Pulmonary Hypertension
Background— Because of high pulmonary vascular resistance in patients with primary pulmonary hypertension (PPH), right atrial pressure may exceed left atrial pressure during exercise, resulting in a right-to-left shunt via a patent foramen ovale (PFO). This shunting would disturb arterial Pco2 and H+ homeostasis if the pulmonary blood were not simultaneously hyperventilated to compensate for the high CO2 and H+ in the shunted blood. This article first hypothesizes and then describes unique changes in gas exchange when right-to-left exercise-induced shunting (EIS) occurs.
Methods and Results— Retrospectively, the cardiopulmonary exercise tests of 71 PPH patients were studied. Criteria postulated to document hyperventilation of the pulmonary blood flow due to a right-to-left EIS were (1) an abrupt and sustained increase in end-tidal O2 with a simultaneous sustained decrease in end-tidal CO2; (2) an abrupt and sustained increase in the respiratory exchange ratio; and (3) usually, an associated decline in pulse oximetry saturation. Each patient was evaluated for a PFO with resting echocardiography. The investigators interpreting the gas exchange evidence of EIS were blinded to the echocardiographic readings. Forty-five percent of the patients had demonstrable EIS by gas exchange criteria. Almost all were also positive for a PFO by echocardiography. Using the resting echocardiograph as the reference, the sensitivity, specificity, positive and negative predictive values, and accuracy were all between 90% to 96%.
Conclusions— Exercise-induced right-to-left shunting can be detected by noninvasive, cardiopulmonary exercise testing in patients with PPH.
Received August 17, 2001; revision received October 24, 2001; accepted October 25, 2001.
Patients with right-to-left intracardiac shunts regulate arterial Pco2 and pH by hyperventilating unshunted lung blood flow to compensate for the high CO2 content in shunted blood.1 During clinical cardiopulmonary exercise testing (CPET) of patients with primary pulmonary hypertension (PPH), we frequently observed gas exchange patterns indicating acute hyperventilation of pulmonary blood flow, suggesting shunting via a patent foramen ovale (PFO).
In normal persons, right-to-left shunting via a PFO is unlikely because left atrial pressure exceeds right atrial pressure. However, with abnormally high pulmonary vascular resistance (as in PPH), right atrial pressure can exceed left atrial pressure, especially during exercise, and force venous (low Po2 and high Pco2 and H+) blood through a PFO directly into the systemic circulation, stimulating systemic arterial chemoreceptors and causing hyperventilation of the unshunted pulmonary blood flow. This compensatory hyperventilation increases CO2 unloading, thereby maintaining arterial Pco2 and H+ homeostasis, despite the presence of an exercise-induced right-to-left shunt (EIS).
The objective of the present study was to describe the specific gas exchange changes that can be used to identify those patients with PPH who develop an EIS.
The Institutional Review Board at Harbor-UCLA approved this protocol. We evaluated the clinical records and gas exchange evidence for or against EIS in 71 PPH patients2 and compared these determinations, in a blinded fashion, with the evidence for PFO found during resting echocardiography. The majority of patients were in New York Heart Association (NYHA) class III. None had other significant concurrent disorders.
CPET Protocol, Measurements, and Data Display
Each patient performed standard, physician-supervised, progressively increasing work rate (WR) CPET on an electromagnetically braked cycle ergometer.2,3 Breath-by-breath gas exchange measurements (Cardiorespiratory Diagnosis System, Medical Graphics), and recordings of ECG, systemic blood pressure, and pulse oximeter saturation (Spo2) were made during 3 minutes of rest, 3 minutes of unloaded cycling, progressively increasing WR exercise, and 2 minutes of recovery.2–4 WR was increased at 5 to 15 W/min (mean, 10±3 W/min), depending on the physician evaluation of patient fitness. Data were interpolated second-by-second, averaged every 10 seconds, and then printed in a table and on a single-page 9-panel plot with 15 graphic displays of the patient’s responses to CPET (Figure 1).2–6 For this investigation, only the first CPET of each patient was analyzed.
Concurrently, CPET studies were obtained from a control group of 20 normal subjects with a similar age, sex, and body size.
Detection of EIS by Gas Exchange Criteria
Three author-investigators (graders), who were blinded regarding all identifying patient information and echocardiographic findings, independently reviewed the 9-panel CPET plots using the following criteria to identify an EIS at the start of unloaded cycling exercise: (1) an abrupt and sustained increase in end-tidal O2 (Peto2), with a simultaneous, sustained decrease in end-tidal CO2 (Petco2) (Figure 1, panel 9); (2) an abrupt and sustained increase in the respiratory exchange ratio (RER=carbon dioxide elimination/oxygen consumption [V̇co2/V̇o2]; Figure 1, panel 8), and (3) usually, an associated Spo2 decline (Figure 1, panel 9).
All patients underwent resting transthoracic echocardiography with Valsalva maneuvers and bubble studies.7–10 The great majority had >1 echocardiogram on different days. In addition, ≈1 of 10 had transesophageal echocardiography. If an atrial right-to-left shunt was detected during any echocardiographic study, the patient was categorized as PFO-positive (PFO+); if not, the patient was categorized as PFO-negative (PFO−).
Separation of PPH Patients into Groups
Using the above criteria applied to the 9-panel CPET graphic array, the 71 CPET studies were independently graded as either EIS-positive (EIS+) or EIS-negative (EIS−) by 3 graders. Two graders also used tabular data to aid in their decision-making when the changes in the graphic data were less obvious. Patients who were PFO+ by echocardiography and unanimously EIS+ were placed in the shunt group; those who were PFO− by echocardiography and unanimously EIS− were placed in the no-shunt group. Any PFO+ patient categorized as EIS− or any PFO− patient categorized as EIS+ by any grader was placed in the discordant group. During this process, all 3 graders independently identified 3 patients who were EIS− during unloaded cycling but converted to EIS+ near the end of their CPET; these were excluded from the grouping.
Data are expressed as mean±SD, except where specifically noted. Most CPET values are expressed as a percent of predicted value.2,3,11,12 Repeated ANOVA with 2-tailed Scheffe tests were used to identify differences between groups; paired t tests were used to identify changes from rest.13,14 P<0.05 was considered significant. Sensitivity, specificity, and predictive values of EIS detection of shunt were calculated,14 despite knowing that a PFO induced during exercise might be unrecognizable during resting echocardiography.
Similarities of PPH Groups at Rest and Peak Exercise
All CPET studies were completed without adverse events. The demographics of the shunt, no-shunt, discordant, and control groups were similar (Table 1). Except for a higher ventilatory equivalent for CO2 (V̇e/V̇co2) at the anaerobic threshold and a higher slope of V̇e versus V̇co2, all of the peak exercise CPET findings of the 3 PPH patient groups were similar to each other (Table 1) but dissimilar from findings in the control group.
Differences in Gas Exchange Between Shunt and No-Shunt Groups
Figure 2 contrasts key CPET measurements that distinguish 2 representative PPH patients (one EIS+ and one EIS−) from a normal subject. Figure 3 describes the second-by-second mean values at rest and during the 3 minutes of unloaded cycling exercise for the same variables in the shunt, no-shunt, and control groups. In the PPH groups, gas exchange was impaired at rest (low Petco2 with high ventilatory equivalent for O2 [V̇e/V̇o2], V̇e/V̇co2, and Peto2), with the Petco2 lowest in the shunt group (Figures 2 and 3⇓). After beginning unloaded cycling, the shunt group abruptly decreased their Petco2, while the Peto2, V̇e/V̇o2, and RER concurrently abruptly increased (Figures 2 and 3⇓). Shortly thereafter, the Spo2 declined in most of the shunt patients. In contrast to the shunt group, the no-shunt group showed lesser changes in Peto2, Petco2, RER, V̇e/V̇o2, and Spo2.
Table 2 summarizes the changes in Peto2, Petco2, V̇e/V̇o2, V̇e/V̇co2, RER, V̇e, and Spo2 from rest to the end of unloaded cycling that distinguish the shunt group from the no-shunt group and the statistical significance of these changes. However, by the end of unloaded cycling, all groups had normal and similar increases in V̇o2 and V̇co2.
The Late-Developing EIS
Figure 4 depicts 1 of 3 patients who developed unmistakable gas exchange evidence of a late-developing EIS during CPET, just before stopping. As with an EIS during unloaded cycling, a late-developing EIS is characterized by abrupt and striking decreases in Petco2 and Spo2, with concurrent striking increases in Peto2, RER, and V̇e/V̇o2 more than V̇e/V̇co2. This patient was PFO− on repeated echocardiography. Her CPET pattern persisted until 2 years after starting epoprostenol therapy, at which time no further CPET evidence of EIS was noted, reflecting her improvement. The second patient had a similar late-developing EIS, but was PFO− by echocardiography at the time of CPET. Two years previously, before treatment, she had been PFO+ by echocardiography. The third patient with CPET changes typical of a late-developing EIS was PFO− on repeated echocardiography.
Grouping of PPH Patients
Excluding the 3 patients with a late-developing EIS, Figure 5 shows the distribution of the 68 patients among the shunt (n=18) and no-shunt (n=39) groups (all 3 graders agreed) and the discordant group (n=11).
If the resting echocardiograph was used as a reference for PFO detection, the overall sensitivity and specificity for CPET EIS detection would be 90% and 96%, respectively (Table 3). Overall, PFO+ PPH patients would also be CPET EIS+ 94% of the time, whereas PFO− PPH patients would be EIS− 95% of the time. Within the discordant group, echocardiography documented a PFO in 6 patients (Figure 5), but 2 of them had been PFO− on one or more other echocardiographic studies, illustrating the inconstant nature of shunting, even at rest.
Considering all 71 PPH patients, 18 were early EIS+ (by all 3 graders) and PFO+, 6 others were PFO+ and EIS+ by the evaluations of 1 or 2 graders (Figure 5), and 3 others were late EIS+ by all 3 graders. Thus, 38% [(18+6+3=21)/71=38%] had convincing evidence for right-to-left shunting during CPET. Five others (Figure 5), although PFO−, had evidence suggesting that they were EIS+. Thus, up to 45% ([27+5=32]/71=45%) of our patients may have had right-to-left shunting during exercise.
In the presence of a PFO and increased pulmonary vascular resistance, exercise-induced increases in venous return cause right atrial pressure to increase. When it exceeds left atrial pressure, venous return can shunt through a PFO, diverting deoxygenated, acidemic, CO2-rich blood to the systemic circulation. This stimulates arterial chemoreceptors (carotid bodies) to maintain arterial H+ and Paco2 homeostasis, causing an immediate increase in ventilation, as manifested by rapid increases in alveolar Po2 (reflected in a Peto2 increase) and decreases in alveolar Pco2 (reflected in a Petco2 decrease) (Figures 1 through 4⇑⇑⇑). Consequently, CO2 unloading from the unshunted pulmonary blood flow increases as alveolar Pco2 falls, but O2 loading increases less because pulmonary capillary Po2 reaches the flat part of the oxyhemoglobin dissociation curve. Thus, ventilation increases more steeply relative to V̇o2 than V̇co2, resulting in a greater increase in V̇e/V̇o2 than V̇e/V̇co2 and a stepwise increase in RER (Figures 1 through 4⇑⇑⇑).
Even at rest, there are distinctly more gas exchange abnormalities (higher V̇e/V̇o2, V̇e/V̇co2, and Peto2 and lower Petco2 and Spo2) in the shunt than no-shunt groups (Figures 2 and 3⇑). These pre-exercise abnormalities can be attributed to hypoperfusion of well-ventilated lung and probable chronic hyperventilation.2,15
With unloaded cycling, the group differences become more obvious (Figures 2 and 3⇑ and Table 2). In the shunt group, V̇e/V̇o2, Peto2, and RER all increased and Petco2 decreased (indicating an acute ventilatory increase disproportionate to metabolism), and Spo2 decreased. In the no-shunt group, Spo2 declined slightly. In contrast, V̇e/V̇o2 and Peto2 decreased and Petco2 increased in the control group. An abrupt increase in V̇e/V̇co2 always indicated an EIS in our study.
Figure 4 illustrates the concurrent, dramatic, unambiguous gas exchange findings seen when shunting abruptly begins and ends at the end of exercise rather than earlier. Near the end of exercise, the stimuli to the chemoreceptors (and oximeter probe) are robust because the shunted mixed-venous blood is more acidemic, hypercarbic, and hypoxemic; thus, it more strikingly alters ventilation to maintain arterial homeostasis.1
Pitfalls in Detection of a PFO and EIS
Because a PFO may be so small or the interatrial pressure differences so trivial, shunt blood flow may not be demonstrable by echocardiography, even with Valsalva maneuvers.8–10
During CPET, the shunt fraction may be so small or the data so noisy that interpretation of EIS criteria are ambiguous. Other potential problems for clinicians using CPET to detect an EIS include a delayed or imperfect response of the Spo2 signal, a poorly scaled graphic data display, or a too-brief period of pre-exercise CPET data.
Two exercise-induced conditions that might lead the clinician to identify an EIS incorrectly are anxiety-induced hyperventilation or a very low anaerobic threshold .
Acute hyperventilation decreases Petco2 while increasing Peto2, V̇e/V̇o2, V̇e/V̇co2, and RER. However, hyperventilation without shunting of venous CO2 into the systemic circulation is rarely sustained with a stable RER for more than a minute or two during exercise because these patients become CO2-unloaded and acutely alkalemic. With hyperventilation and no other disease, Spo2 does not decrease and MRT for V̇o2 is normal, in contrast to PPH.2,3
With a low anaerobic threshold , the development of lactic acidosis at a low WR causes Peto2, V̇e/V̇o2, and RER to continue to increase, in contrast to the abrupt but stable increases seen with an EIS. Evidence that the anaerobic threshold is reached later during exercise confirms that the earlier changes are due to an EIS.
Validity of Patient Groupings
As demonstrated in Figures 4 and 5⇑, the absence of a detectable PFO at rest does not preclude right-to-left shunting during exercise. Because shunting is a dynamic process dependent on transient pressure differentials, we did not expect to find an absolute concordance between the resting echocardiograph and CPET evidence of an EIS. It is unlikely that even the most sensitive echocardiographic methods at rest would detect PFOs in the 39 patients who were graded as EIS− or that the 18 patients who had a detectable PFO by echocardiography would not also have right-to-left shunting during exercise. Therefore, we used both CPET and echocardiographic findings to define and compare the shunt and no-shunt groups.
Incidence of Right-to-Left Shunting
The high sensitivity, specificity, positive and negative predictive values, and accuracy comparing CPET with echocardiography (Table 3) demonstrate the utility of the gas exchange method in EIS detection. With respect to intraobserver variability, it seems that using tabular data to detect small changes increased sensitivity but slightly decreased specificity.
Ultimately we found, using both the resting and exercise measurements of our 71 PPH patients, that in addition to the 18 patients who were both PFO+ and EIS+ by all graders, 9 more patients (6 by echocardiography and 3 by distinctive late-exercise changes) had convincing evidence of a right-to-left shunt either at rest or during exercise (Figures 4 and 5⇑). Five more may have had right-to-left shunting by CPET criteria (Figure 5). Thus, the incidence of right-to-left shunting through a PFO in our PPH patients during exercise seems to be 38% to 45%.
An autopsy study of 965 “normal” hearts showed a PFO incidence of 20% to 34%, with decreasing PFO frequency but increasing size with advancing age.16 Using Valsalva maneuvers during echocardiography, the incidence of PFO in normal subjects is reported at just 5% to 18%,7–10 in part because most normal adults do not shunt blood through their PFO and also because the Valsalva maneuver does not always produce sufficient interatrial pressure differences to cause shunting. An 18% echocardiographic incidence of PFO or interatrial defects was detected in a recent series of untreated PPH patients.17 Stroke patients have a higher incidence of PFO (as high as 78% in young patients with cryptogenic strokes, possibly due to paradoxical emboli).8–10 The relatively high EIS+ incidence in our series argues that chronic pulmonary hypertension also increases the potential for shunting through foramina ovale that might otherwise remain undetected and that such shunting may favor an increased survival.
CPET is a safe, noninvasive, cost-effective, and easily repeatable method for assessing PPH patients2 and detecting an EIS. The 9-panel graphic array (Figure 1) not only helps in the general interpretation of CPET studies, but also assists in the recognition of the distinctive gas exchange pattern of an EIS.
Sietsema KE, Cooper DM, Perloff JK, et al. Control of ventilation during exercise in patients with central venous-to-systemic arterial shunts. J Appl Physiol. 1988; 64: 234–242.
Sun XG, Hansen JE, Oudiz RJ, Wasserman K. Exercise pathophysiology in patients with primary pulmonary hypertension. Circulation. 2001; 104: 429–435.
Wasserman K, Hansen JE, Sue DY, et al. Principles of Exercise Testing and Interpretation. 3rd ed. Baltimore: Lippincott Williams & Wilkins; 1999.
Beaver WL, Wasserman K, Whipp BJ. On-line computer analysis and breath-by-breath graphical display of exercise function tests. J Appl Physiol. 1973; 34: 128–132.
Sue DY, Hansen JE, Blais M, Wasserman K. Measurements and analysis of gas exchange during a programmable calculator. J Appl Physiol. 1980; 49: 456–461.
Sietsema KE, Cooper DM, Perloff JK, et al. Dynamics of oxygen uptake during exercise in adults with cyanotic congenital heart disease. Circulation. 1986; 73: 1137–1144.
Serena J, Segura T, Perez-Ayuso MJ, et al. The need to quantify right-to-left shunt in acute ischemic stroke. Stroke. 1998; 29: 1322–1328.
Habedank D, Reindl I, Vietzke G, et al. Ventilatory efficiency and exercise tolerance in 101 healthy volunteers. Eur J Appl Physiol. 1998; 77: 421–426.
Kleber FX, Vietzke G, Wernecke KD, et al. Impairment of ventilatory efficiency in heart failure. Prognostic impact. Circulation. 2000; 101: 2803–2809.
Glantz SA. Primer of Biostatistics. 4th ed. San Francisco: McGraw-Hill; 1997.
Portney LG, Watkins MP. Foundations of Clinical Research: Applications to Practice. 2nd ed. Upper Saddle River: Prentice-Hall; 2000.
Paciocco G, Martinez FJ, Bossone E, et al. Oxygen desaturation on the six-minute walk test an mortality in untreated primary pulmonary hypertension. Eur Resp J. 2001; 17: 647–652.