In this study, we tested patients with acute respiratory failure for the presence of a patent foramen ovale using saline contrast transesophageal echocardiography. All patients were receiving mechanical ventilation, and we compared systemic oxygenation and right-to-left shunt fraction in each patient with and without PEEP. Because PEEP may worsen systemic oxygenation through depression of cardiac output [16, 17], we concurrently measured mixed venous oxygen saturation levels and obtained estimated cardiac output [18]. We report the effects of PEEP in mechanically ventilated patients with a patent foramen ovale and compare them with similar patients without a patent foramen ovale.

Methods

Top Methods Results Discussion Author & Article Info References

Patients

The study was approved by the University of Saskatchewan Ethics Committee on Human Experimentation. Informed consent from the patient or family was obtained. Hemodynamically stable patients in acute respiratory failure who required mechanical ventilation with an inspired oxygen concentration of 50% or more and a PEEP of 5 cm of H₂O or more were considered eligible for the study. Between October 1991 and September 1992, 46 patients were entered into the study; they were selected from 940 of the total admissions to a combined medical-surgical intensive care unit of a university teaching hospital. Specific causes of respiratory failure and clinically significant chest radiographic findings are listed in Table 1. We collected demographic data that included age, sex, weight, medical diagnosis, and Apache II score [19].

Baseline Hemodynamics and Gas Exchange

Patients were hemodynamically stable for at least 1 hour before starting the study. We then obtained baseline measurements of central venous pressure and mean systemic arterial pressure in all patients and, in the 23 patients with a pulmonary artery catheter (Model 93A-1317F; Baxter-Edwards, Santa Ana, California), we also measured pulmonary artery systolic pressure, pulmonary artery diastolic pressure, pulmonary capillary wedge pressure, and cardiac output in triplicate using thermodilution (Merlin Monitor, model 565; Hewlett Packard, Boston, Massachusetts). Arterial and mixed venous blood samples were collected, and we measured the respective blood gas tensions (corrected for temperature) [20] using appropriately calibrated electrodes (Nova Biomedical #75SP1070, Waltham, Massachusetts) as well as the respective oxyhemoglobin saturation levels using co-oximetry (OSM-2 Co-oximeter; Radiograph, Copenhagen, Denmark).

Experimental Protocol

A transesophageal ultrasound transducer (biplane 5 mHz phased-array transducer; Aloka 870, Tokyo, Japan) was inserted into the esophagus after premedication (if necessary) using 2 to 5 mg of midazolam (Hoffman-LaRoche, Mississauga, Ontario) intravenously and 2% lidocaine topical spray. The inspired oxygen concentration was increased to 100% and the level of the PEEP was randomized to 0 cm of H₂O or 10 cm of H₂O. After waiting 15 minutes, all baseline measurements were repeated.

Standard transesophageal images were obtained [21]. Left ventricular internal dimensions and area during systole and diastole were obtained from the transgastric short-axis view [22]. The right ventricular internal dimension was measured during diastole from the four-chamber view at the tricuspid chordal level. Atrial septal motion was assessed in the four-chamber view. Loss of phasic motion with sustained convexity toward the left atrium or the right atrium was described as deviation into the left atrium or the right atrium, respectively [23]. Regurgitation of the tricuspid, mitral, and aortic valves was assessed using color flow imaging. Agitated saline contrast [13] was injected rapidly into the right atrium through the central line or proximal port of the pulmonary artery catheter for detection of patent foramen ovale. Transesophageal images were obtained in either the four-chamber view or the long-axis ascending aorta-atrial septal view [21] during the injection of saline contrast. Each saline injection was repeated no less than three times and was determined to be adequate if the right atrium was entirely opacified with microbubbles. A patent foramen ovale was diagnosed if, on any injection, three or more bubbles were visualized in the left atrium within three cardiac cycles from the time of complete opacification of the right atrium [13, 23, 24]. Semiquantification of the right-to-left intracardiac shunt was done using the following criteria: 1) a small shunt was less than 5 microbubbles in the left atrium on any frozen frame; 2) a moderate shunt was 5 to 20 microbubbles in the left atrium on any frozen frame; 3) a severe shunt was more than 20 microbubbles or more than 50% of the left atrium was opacified on any frozen frame. All agitated saline injections were assessed by two experienced observers, and injections were repeated until agreement was reached on the interpretation of the study.

After these measurements, the alternate level of PEEP was set, and all hemodynamic, blood gas, and echocardiographic measurements were repeated (for example, those patients initially studied using a PEEP of 10 cm of H₂O were switched to 0 cm of H₂O and those studied using a PEEP of 0 cm of H₂O were switched to 10 cm of H₂O).

Calculations

The total right-to-left shunt fraction (Qs/Qt) was calculated using the shunt equation

Qs/Qt = (Cc_O2 – Ca_O2)/

(Cc_O2 – Cv_O2)

where CaO₂ (arterial oxygen content) and CvO₂ (mixed venous oxygen content) were calculated from the directly measured hemoglobin and blood oxyhemoglobin saturation levels [25]. The end-capillary oxygen content (CcO₂) and the alveolar-to-arterial difference for oxygen were calculated using the alveolar gas equation.

The left ventricular fractional area change was defined as the left ventricular area during diastole minus the left ventricular area during systole divided by the left ventricular area during diastole [22]. The stroke volume was calculated using left ventricular volumes obtained by the Teichholz method [18, 26] as follows:

stroke volume = (7/[2.4 + diastolic dimension] x diastolic dimension³) minus (7/[2.4 + systolic dimension] x systolic dimension³).

The cardiac index was then calculated by multiplying the stroke volume by the heart rate and dividing by the body surface area.

Statistics

Hemodynamic, gas exchange, and echocardiographic data between patients were compared by paired t-tests. These data were also compared between groups by unpaired t-tests. Percentages in different groups were compared using a chi-square test with Yates correction or the Fisher exact test. Linear regression was calculated using the method of least squares. Statistical significance was defined as P < 0.05 [27]. Measurements are expressed as mean ± standard deviation.

Results

Top Methods Results Discussion Author & Article Info References

Demographic and Baseline Characteristics

Forty-six patients were recruited and 7 (15%) were found to have a patent foramen ovale by saline contrast transesophageal echocardiography. No statistical differences were noted between the 39 patients without a patent foramen ovale and the 7 patients with a patent foramen ovale in terms of age (58 ± 19 years compared with 65 ± 11 years), sex (12 women; 27 men compared with 3 women; 4 men), days to study after onset of respiratory failure (2.3 ± 2.9 days compared with 1.6 ± 1.1 days), and Apache II score (20 ± 5 compared with 20 ± 7).

Before insertion of the ultrasound transducer, no differences were noted between patients without or with a patent foramen ovale in terms of inspired oxygen concentration (61% ± 11% compared with 60% ± 10%), PEEP (7.1 ± 2.8 cm of H₂O compared with 6.8 ± 2.4 cm H₂O), pH (7.42 ± 0.07 compared with 7.40 ± 0.07), PCO₂ (24 ± 7 mm Hg compared with 36 ± 7 mm Hg), PO₂ (86 ± 25 mm Hg compared with 74 ± 16 mm Hg), the alveolar-to-arterial difference for oxygen (275 ± 81 mm Hg compared with 281 ± 66 mm Hg), the mean systemic blood pressure (81 ± 13 mm Hg compared with 74 ± 16 mm Hg), central venous pressure (12 ± 6 mm Hg compared with 9 ± 6 mm Hg), and diffuse chest radiographic findings (31 of 39 compared with 5 of 7 patients, P > 0.2) or regional chest radiographic findings (8 of 39 compared with 2 of 7 patients, P > 0.2).

Other Cardiac Diseases

Left ventricular systolic function was assessed by a visual estimate of the left ventricular ejection fraction (LVEF). Of the 7 patients with patent foramen ovale, 5 (71%) patients had normal left ventricular systolic function (LVEF > 60%), 1 (14%) patient had mild systolic dysfunction (LVEF between 40% and 60%), and 1 (14%) patient had moderate-severe left ventricular dysfunction (LVEF < 40%). No patient with a patent foramen ovale had more than mild tricuspid regurgitation or greater than 2+ regurgitation of the left heart valves. Of the 39 patients without a patent foramen ovale, 22 (56%) patients had normal left ventricular function, 4 (10%) patients had mild left ventricular systolic dysfunction, 10 (26%) patients had moderate-severe left ventricular systolic dysfunction, and 2 (5%) patients had left ventricular diastolic dysfunction (left atrial enlargement, spontaneous echo contrast in the left atrium, and left ventricular hypertrophy). One (3%) patient had severe aortic regurgitation. Left ventricular function was similar when comparing those patients with patent foramen ovale with those without patent foramen ovale.

Effect of Positive End-Expiratory Pressure

Table 2 contains hemodynamic findings and blood gas levels in those patients with a patent foramen ovale and in those without a patent foramen ovale using PEEP (0 cm of H₂O and 10 cm of H₂O). In patients without a patent foramen ovale, increasing the PEEP from 0 cm of H₂O to 10 cm of H₂O was associated with increased arterial PO₂ (P < 0.001), increased arterial oxyhemoglobin saturation (P < 0.01), decreased alveolar-to-arterial difference for oxygen ( –50 mm Hg; CI, –21 to –67;P < 0.001), and decreased shunt fraction ( –0.05;CI, –0.03 to –0.07;P < 0.001). Neither the cardiac index by echocardiography, the mixed venous oxyhemoglobin saturation level, nor the mixed venous PO₂ value were statistically different after increasing the PEEP. Twenty of these patients had pulmonary artery catheters in place before beginning the study, and in them the cardiac index measured by thermodilution did not change with the addition of PEEP.

In the seven patients with a patent foramen ovale, the addition of PEEP (10 cm of H₂O) did not alter the arterial PO₂, the alveolar-to-arterial difference for oxygen, or the cardiac index. Similarly, neither the mixed venous oxyhemoglobin saturation level nor the mixed venous PO₂ value changed when the PEEP was increased. However, the total shunt fraction did increase (0.05; CI, 0 to 0.09; P < 0.05) when the PEEP was increased, from 0 cm of H₂O to 10 cm of H₂O.

In all 46 patients, the addition of PEEP did not alter the central venous pressure, the heart rate, or the mean systemic blood pressure. Similarly, in those 23 patients with a pulmonary artery catheter in place, the values of pulmonary artery systolic pressure, pulmonary artery diastolic pressure, and pulmonary capillary wedge pressure did not change with the addition of PEEP.

Figure 1 illustrates the change in the alveolar-to-arterial difference for oxygen and the change in the shunt fraction after increasing the PEEP from 0 cm of H₂O to 10 cm of H₂O. These changes are shown for the patients with and without a patent foramen ovale. The changes in the shunt fraction (P = 0.003) and the alveolar-to-arterial difference for oxygen (P = 0.044) after the addition of PEEP were different after comparing patients with and without a patent foramen ovale.

View larger version (19K): [in this window] [in a new window]   Figure 1. The changes after adding positive end-expiratory pressure (10 cm of H2O) in patients with and without a patent foramen ovale. Changes in the alveolar-to-arterial difference for oxygen (left) and in shunting (right) are shown in those patients with a patent foramen ovale (hatched bar) and in those patients without a patent foramen ovale (white bar). Note that increases in shunting and increases in the alveolar-to-arterial difference for oxygen are represented by negative values and reflect more severe conditions.* P = 0.003;** P = 0.044.

Transesophageal Echocardiography

Echocardiographic measurements were obtained in all patients. The right ventricular internal dimension (range, 26 to 39 mm) and the left ventricular fractional area change (range, 22% to 62%) were not changed after the addition of PEEP in patients with or without a patent foramen ovale. The proportion of patients with a deviation of the atrial septum into the left atrium was greater (P < 0.01) in those patients with a patent foramen ovale (5 of 7 patients, 72%) compared with patients without a patent foramen ovale (6 of 39 patients, 15%).

Three of seven patients with patent foramen ovale had a small right-to-left shunt as assessed by contrast echocardiography. Three patients had moderate right-to-left shunting and one patient had severe right-to-left shunting. The addition of PEEP (10 cm of H₂O) caused an increase in the echocardiographic grading of the right-to-left shunt from small to moderate in only one patient. Another patient had no shunting with a PEEP of 10 cm of H₂O and had a small shunt with a PEEP of 0 cm of H₂O. This patient had deviation of the interatrial septum into the right atrium suggestive of increased left atrial pressure. This would limit the severity of right-to-left intracardiac shunting.

Shunt Changes

Thirteen patients had an increase in shunting with the addition of a PEEP of 10 cm of H₂O. Of these 13 patients, 6 patients had a patent foramen ovale and 7 did not have a patent foramen ovale. Table 3 illustrates the changes in blood gases and hemodynamic variables in the 39 patients without a patent foramen ovale; the 7 patients who had increased shunting in response to increasing the PEEP were analyzed separately from the remaining 32 patients who had decreased shunting in response to increasing the PEEP. In those patients in whom shunting decreased after increasing the PEEP, the levels of associated arterial PO₂ increased (P < 0.001), the levels of arterial oxyhemoglobin saturation increased (P < 0.02), and the alveolar-to-arterial difference for oxygen decreased (P < 0.001). In those patients in whom shunting increased (P < 0.002) after increasing the PEEP, no associated changes were noted in values of arterial PO₂, arterial oxyhemoglobin saturation, or alveolar-to-arterial difference for oxygen. Central venous pressure and echocardiographic measurements of right ventricular dimensions and left ventricular fractional area change were similar in those patients who had increased shunting compared with those who had decreased shunting in response to the addition of PEEP. Regional or diffuse changes on chest radiograph were not statistically different (P > 0.2) when comparing patients who did and did not respond to PEEP.

Discussion

Top Methods Results Discussion Author & Article Info References

Prevalence

The prevalence of patent foramen ovale in our study was 15%, which is slightly lower than that found at autopsy, for which the prevalence ranged from 34.3% during the first three decades of life to 20.2% during the 9th and 10th decades [9]. Contrast transesophageal echocardiography can be used to detect patent foramen ovale during life [13], but this requires the presence of a positive pressure gradient from the right-to-left atrium to produce a right-to-left shunt. A transient increase in right atrial pressure above left atrial pressure occurs with each cardiac cycle and can be accentuated by respiratory maneuvers [28]. In spontaneously breathing patients, the Valsalva maneuver during transesophageal contrast echocardiography increases the prevalence of patent foramen ovale from 12% to 20% [24]. During mechanical ventilation, holding peak inspiratory pressure at 20 cm of H₂O [29] or adding PEEP [3, 10] has been used to enhance right-to-left shunting and thereby increase the detection of patent foramen ovale by contrast echocardiography. The high peak inspiratory pressures of the patients with acute respiratory failure in our study probably increased the sensitivity of our method of detecting a patent foramen ovale. Six of seven patients with patent foramen ovale had right-to-left shunting at the atrial level with or without PEEP. However, we still found a slightly lower prevalence of patent foramen ovale in our study when compared with other series in normal persons [12] or in patients having elective operative procedures [29] despite using identical technology. This may represent a selection bias of our study, because patients with severe right-to-left shunting through a patent foramen ovale may not have survived the superimposed acute respiratory failure or may have hemodynamic instability preventing entry into the study. Thus, we may have underestimated the prevalence of patent foramen ovale in the intensive care population.

Shunting

The 46 patients in our study had severe respiratory failure with total right-to-left shunts of 30% and alveolar-to-arterial oxygen differences of 500 mm Hg on 100% oxygen. Because all patients received 100% oxygen during the study, none of them became clinically hypoxemic. Mean arterial oxygenation statistically increased with the increase in PEEP in the group of patients without a patent foramen ovale. However, in the seven patients with a patent foramen ovale, increasing the PEEP was associated with increased shunting and no improvement in arterial oxygenation. By breathing 100% oxygen, the effect of ventilation-to-perfusion mismatch on gas exchange is eliminated, as are variations due to air/oxygen mixing [25]. The only cause of hypoxemia that remains is that due to right-to-left shunting. There are essentially three sources of shunting that would be measured in this study. A small physiologic shunt exists that is related to thebesian and bronchial vein drainage. Of greater clinical importance is intracardiac shunting due to flow of deoxygenated blood from a right-to-left heart chamber and, finally, intrapulmonary shunting due to perfusion of nonventilated lung units. Contrast transesophageal echocardiography can determine the presence or absence of an intracardiac shunt but does not allow the precise quantitation of the amount of intracardiac shunting. Qualitatively, we detected a difference in shunting using a PEEP between 0 cm of H₂₀ and 10 cm of H₂₀ in only one of seven patients with patent foramen ovale. However, in the absence of a patent foramen ovale, we found that adding PEEP (10 cm of H₂O) decreased the total shunt fraction from 30% to 25%, whereas in the presence of a patent foramen ovale, the total shunt fraction increased from 28% to 32%. We can, therefore, estimate that the addition of PEEP, in patients with patent foramen ovale, increased the intracardiac shunt by 9% (for example, the intrapulmonary shunt decreased by 5% and the intracardiac shunt increased by 9% to result in a net increase in total shunting of 4%).

Limitations

One limitation of this study was our inability to obtain blood samples from the pulmonary artery in all patients. We measured mixed venous oxygen from a blood sample withdrawn from the central line rather than from the pulmonary artery, because not all patients had a pulmonary artery catheter in place. Although mixing in the right atrium may not be complete, central venous line and pulmonary artery sampling for oxygen concentration have been shown to be correlated [30]. In the 23 patients with a pulmonary artery catheter, we measured mixed venous PO₂ and mixed venous saturation in blood samples drawn almost simultaneously from the right atrium and the pulmonary artery. Mixed venous PO₂ at the two sites were correlated (r² = 0.65), as were the mixed venous oxyhemoglobin saturation levels at these sites (r² = 0.67). Therefore, we were confident that measurements of right atrial PO₂ reflected changes in mixed venous PO₂ and did not influence our measurement of the intrapulmonary shunt [31]. In addition, we did not measure cardiac output by thermodilution but rather by echocardiographically determined changes in left ventricular systolic and diastolic chamber dimensions [18]. In the 23 patients with a pulmonary artery catheter in place, cardiac output measured by thermodilution correlated well with cardiac output measured echocardiographically (P < 0.018). In clinical practice, thermodilution cardiac output determinations can at best detect differences of approximately 13% ± 3% [32]. Changing the intracardiac shunt would make these differences even less reliable. Using either technique, we did not detect any changes in cardiac output after the addition of PEEP.

Effects of Positive End-Expiratory Pressure

Although PEEP is used in mechanically ventilated patients to improve systemic oxygenation and decrease high inspired oxygen concentrations [14], it is not effective in all patients. The cardiopulmonary effects of PEEP may have conflicting effects on systemic oxygenation. Positive end-expiratory pressure may decrease cardiac output by decreasing systemic venous return [33] or by altering the ventricular pressure-volume relation [34]. The decrease in cardiac output may decrease mixed venous oxygen saturation, which in turn worsens systemic arterial oxygenation [14, 17]. Concurrently, as mixed venous oxygen saturation decreases, intrapulmonary shunting decreases, which has the effect of increasing systemic oxygenation [31, 35]. Positive end-expiratory pressure may also worsen systemic oxygenation, particularly in the setting of regional lung disease; PEEP improves gas exchange primarily through redistribution of pulmonary edema out of the alveolar spaces into the interstitial spaces [36]. However, if the alveolar exudate cannot be readily translocated to the interstitium, PEEP may decrease the regional perfusion of the relatively normal lung units and, if cardiac output remains unchanged, PEEP may increase shunting through the nonventilated units [37]. Finally, PEEP may worsen systemic oxygenation by inhibiting hypoxic pulmonary vasoconstriction [38]. Augmentation of hypoxic pulmonary vasoconstriction can decrease intrapulmonary shunting [39]; conversely, inhibition of hypoxic pulmonary vasoconstriction would be expected to increase intrapulmonary shunting and thereby worsen systemic oxygenation. We speculate that the effects of PEEP on hypoxic pulmonary vasoconstriction or on regional perfusion worsened right-to-left shunting with the addition of PEEP in 7 of 39 (18%) patients without patent foramen ovale. Because most patients with patent foramen ovale (6 of 7 patients, 86%) had increased right-to-left shunting with PEEP, and because this group of patients did not have a higher prevalence of regional lung disease, intracardiac shunting of deoxygenated blood is more likely to have played a major role in these patients.

We prospectively selected a PEEP level of 10 cm of H₂O, because this would be unlikely to alter cardiac output [14]. The cardiac index may not decrease in hemodynamically stable patients until levels of PEEP reach 15 cm of H₂O or higher [40]. Similarly, canine studies [41] have confirmed that the cardiac index does not decrease at PEEP levels as high as 15 cm of H₂O if the transmural cardiac filling pressures are maintained near normal. We found that only 1 of 46 patients transiently decreased their blood pressure after adding PEEP. We did not find statistically significant changes in either the cardiac index or in mixed venous oxyhemoglobin saturation after adding a PEEP of 10 cm of H₂O. In addition, we did not find changes in indices of left ventricular performance or changes in right ventricular diastolic dimensions after adding PEEP. Thus, no indication was noted that adding PEEP altered either right or left ventricular function in these patients. Therefore, our observed changes in systemic oxygenation or in shunting cannot be explained by the cardiac effects of PEEP.

Therapy to improve oxygenation in a patient with a patent foramen ovale who does not respond to PEEP is not well defined. If the hypoxemia is due to pneumonia, then cyclooxygenase inhibitors could be beneficial [42]. Infusion of prostaglandin E₁ into the pulmonary artery [43] or inhalation of nitric oxide [44] might decrease pulmonary arterial pressure selectively with minimal systemic arterial effects. However, even low concentrations of inhaled nitric oxide may cause a pulmonary injury [45], and the use of either agent in this setting remains speculative. Alternatively, a decrease in inspired pressure during mechanical ventilation would result in a decrease in the transmitted effect on pulmonary artery pressure [14, 16, 34]. Operative or balloon closure of the patent foramen ovale has been done in patients with refractory hypoxemia [4-7].

Overall, PEEP increased right-to-left shunting in 13 of 46 patients (28%) with acute respiratory failure. Nearly 50% (6 of 13) of the patients who increased right-to-left shunting with PEEP had a patent foramen ovale. We conclude that the lack of improvement after adding PEEP in patients with patent foramen ovale may be due to increased intracardiac shunting. Intracardiac shunting through a patent foramen ovale may be of greater clinical significance in patients with acute respiratory failure because of the associated pulmonary hypertension, mechanical ventilation, and PEEP—all of which increase right atrial pressure and thereby increase the pressure gradient driving flow into the left atrium.

Abbreviation

LVEF: left ventricular ejection fraction

PEEP: positive end-expiratory pressure