There is presently no simple noninvasive technique to assess left ventricular (LV) contractility at the bedside. Changes in slope of the LV end-systolic pressure-volume relationship (ESPVR), known as end-systolic elastance (Ees), are sensitive measures of changes in cardiac contractility independent of concomitant changes in LV preload or afterload [1]. However, to measure Ees one must measure both LV pressure and volume while perturbing the cardiovascular system so that a series of end-systolic pressure-volume data pairs are created. Importantly, noninvasive estimates of LV end-systolic volume can be made using echocardiographic imaging techniques [2]. Furthermore, both arterial pressure [3] and noninvasive estimates of arterial pressure [4] can be substituted for LV pressure. Thus the primary limitation to assessment of LV contractility is the rapid change in LV preload. Traditionally, bolus vasodilator infusions (nitroprusside) or transient inferior vena caval (IVC) occlusions are used to produce the needed changes in LV preload over sequential beats. However, nitroprusside may cause clinically significant hypotension and IVC occlusion requires an indwelling IVC balloon catheter. And neither is suitable for repetitive measures of contractility change over time. If a simple, less invasive method existed that predictably decreases LV preload, one could estimate Ees at the bedside and use it to diagnose disease and guide therapy.

Haney et al. [5] suggested that an inspiratory hold maneuver can be used to create transient decreases in LV preload by transiently decreasing the pressure gradient for venous return ultimately decreasing LV preload [6]. Unfortunately, inflation may also increase intra-abdominal pressure and thus prevent any decrease in LV preload [7]. Furthermore, dynamic changes in lung volume induce their own changes in right atrial pressure, pulmonary vascular resistance, and LV filling [8]. Intrathoracic pressure (ITP) must increase for CPAP to reduce LV preload. If ITP increases greatly, it artificially alters the resultant end-systolic pressure-volume domain. However, low levels of CPAP would not induce large ITP increases. Thus if low levels of CPAP could induce changes in venous return, they could be used to assess LV contractility.

We hypothesized that CPAP-induced changes in LV preload would produce a decrease in LV preload allowing estimation of Ees, whose values would change in a similar fashion to Ees values produced by IVC occlusion. However, we also hypothesized that as CPAP level increased this relationship would be lost. Our results, as recently presented [9], suggest that low levels of CPAP can be used to estimate Ees, but that as CPAP increases this relationship is degraded.


Surgical preparation

After approval of the institutional Animal Care and Use Committee we studied 18 mongrel dogs (20.5 ± 1.3 kg body wt). Two animals were omitted from the subsequent analysis because of persistent hemodynamic instability; the data on the remaining 16 dogs are summarized in Table 1. Anesthesia was induced with sodium pentobarbital (30 mg/kg intravenous bolus) and maintained with a continuous infusion (0.1 mg/kg per minute with intermittent boluses as needed) and a maintenance infusion of 0.9% NaCl at 100 ml/h. All animals were orally intubated and their lungs ventilated with room air. Arterial blood gases were measured periodically. Acid-base status was adjusted with intermittent boluses of sodium bicarbonate solution, as needed, to maintain arterial blood pH between 7.35 and 7.45, and the ventilator was adjusted to maintain arterial PCO2 between 35 and 45 mmHg. Body temperature was maintained between 36° and 38°C by using a heating blanket.

Table 1 Mean left ventricular pressure, volumes, and calculated contractile measures (n = 16) during steady state and at end of preload reduction for all methods of acute preload reduction. (IVC inferior vena caval, CPAP continuous positive airway pressure, EDP end-diastolic pressure, EDV end-diastolic volume, ESP end-systolic pressure, ESV end-systolic volume, Ees end-systolic elastance, PRSW preload-recruitable stroke work)

Fluid-filled catheters were inserted into the thoracic aorta and inferior vena cava. A 7.5-F balloon-tipped pulmonary artery catheter (Edwards LifeSciences, Irvine, Calif., USA) was placed to measure cardiac output. IVC occlusions were performed using an 8-F, 43-ml Fogarty occlusion catheter (Edwards LifeSciences) placed at the level of the supradiaphragmatic IVC. A 5-F, high-fidelity pressure-tipped transducer (Micro-tip catheter MPC-500, Millar, Houston, Tex., USA) was placed in the LV. A second high-fidelity pressure-tipped transducer was placed in the intrathoracic aorta. A 6-F 12-pole, multielectrode conductance catheter (Webster Laboratories, Baldwin Park, Calif., USA) was inserted into the LV to measure LV volume using the method of Baan et al. [10], as previously validated by us [10, 11, 12, 13]. Placement was confirmed by fluoroscopy and inspection of the regional volume-time signals.


The protocol was initiated after at least 30 min of hemodynamic stability, defined as a stable heart rate, blood pressure without evidence of arrhythmias, or metabolic acidosis. For each cardiovascular state we initially reduced LV end-diastolic volume by transient IVC occlusion during a brief (< 5 s) apnea at zero airway pressure until LV end-systolic pressure decreased to below 15 mmHg. Then similarly timed inspiratory hold maneuvers were performed at 5, 10, and 15 mmHg (CPAP). CPAP was induced by hand using a 3-l ventilation bag (Ambubag). Inspiratory hold maneuvers were sustained for 5–10 s. A minimum of 60 s separated each sequential preload-reducing maneuver. After these initial control runs we reduced LV contractility by a continuous esmolol infusion (20 mg intravenous bolus followed by 2 mg/min; Esmolol) followed after washout by increasing contractility by a dobutamine infusion (5 μg/kg per minute; Dobuta). At the conclusion of the study period the dogs were enrolled in unrelated studies. Following all studies the animals were killed using a sequential dose of first pentobarbital (100 mg intravenous bolus) followed by supersaturated potassium chloride (10 ml intravenous bolus).

Data collection

LV pressure, volume, and electrocardiographic signals were digitized at 150 Hz and stored on disk for off-line analysis (Ponemah System, Gould, Cleveland, Ohio, USA). We recorded aortic blood pressure, pulmonary artery pressure, right atrial pressure, and airway pressure (Gould). Left ventricular volume signals were collected and processed using a conductance catheter data processor and signal conditioner (Leycom Sigma 5DF, Leycom Sigma Leyden, The Netherlands). Cardiac output was determined in triplicate using a 10-ml room temperature saline bolus thermodilution technique. LV ESPVR was calculated by the iterative technique using the method of least squares [14]. The slope the ESPVR was taken to reflect Ees, whereas the change in the slope in response to pharmacological intervention was taken to reflect immediate changes in contractility. Preload-recruitable stroke work (PRSW) was also calculated as the slope of the LV stroke work to LV end-diastolic volume relationship during a rapid IVC occlusion maneuver [15]. All results are reported as mean ± SD.

Data analysis

Data were compared among maneuvers and conditions and between animals using a three-way analysis of variance for repeated measures followed by the post hoc Student's Neuman-Keuls test. Significance reports a difference corresponding to a p value less than 0.05. Correlations between estimates of Ees were made using linear regression and Bland-Altman analysis. The effect of CPAP on LV pressure-volume relationships was analyzed by measuring the dynamic increase in LV end-diastolic pressure from immediately before CPAP until Paw increased.


IVC occlusion and CPAP with varying Paw

IVC occlusion and all levels of CPAP consistently decreased LV pressures and volumes, allowing the construction of LV ESPVR and calculation of Ees and PRSW. CPAP induced an immediate increase in LV pressures for the same LV volumes, such that the associated LV ESPVR created was shifted to the left of the apneic LV end-systolic pressure-volume point. The LV diastolic filling relationship was also shifted upward; however, the slope of the LV diastolic filling relationship was not changed (Fig. 1). This increasing LV pressure effect of CPAP increased progressively at CPAP increased from 5 to 10 to 15 mmHg CPAP (Fig. 2).

Fig. 1
figure 1

Left Strip chart recording of airway pressure, left ventricular (LV) pressure, and volume during apnea and with the immediate administration of 5 mmHg CPAP (start CPAP). Right LV pressure-volume relationship of these same data demonstrating the construction of the end-systolic elastance (Ees) relationship created by the CPAP maneuver

Fig. 2
figure 2

LV pressure-volume loops during inferior vena caval (IVC) occlusion and continuous positive airway pressure (CPAP) or 5, 10, and 15 mmHg. Bold lines Apneic baseline and end LV pressure-volume loops for one animal; diagonal lines end-systolic pressure-volume domains for each state. The slope of these lines varies with change sin contractility

Ees was not correlated between the IVC occlusion and CPAP maneuvers (R 2 = 0.43, 0.56, and 0.19 for CPAP 5, 10, and 15 mmHg, respectively) for individual animals. Bias values for Ees between IVC occlusion and CPAP were also large (0.02 ± 4.74, 1.98 ± 5.17, and 5.77 ± 9.74 for CPAP 5, 10, and 15 mmHg, respectively; Fig. 3a). PRSW was correlated more than Ees between IVC occlusion and CPAP (R 2 = 0.43, 0.48, and 0.004 for CPAP 5, 10, and 15 mmHg, respectively) and bias values for PRSW between IVC occlusion and CPAP relatively small (1.58 ± 5.07, 0.06 ± 3.14, and 1.68 ± 4.21 for CPAP 5, 10, and 15 mmHg, respectively; Fig. 3b).

Fig. 3
figure 3

Relationship between end-systolic elastance (Ees; a) and preload-recruitable stroke work (PRSW; b) estimated in the inferior vena caval (IVC) occlusion and continuous positive airway pressure (CPAP). Upper row of each variable Linear regression analysis; lower row Bland-Altman analysis

IVC occlusion and CPAP with esmolol and dobutamine

Esmolol induced a small but significant decrease in heart rate (110 ± 12 vs. 101 ± 8 beats/min, p< 0.05) whereas dobutamine produced a slight and insignificant increase in heart rate (110 ± 12 v. 119 ± 25 beats/min, NS; (Fig. 4 and E1 in the Electronic Supplemental Material). The Ees change across pharmacologically induced changing contractility conditions was similar between IVC occlusion and CPAP maneuvers (Fig. 5). Although esmolol did not consistently decrease Ees, it did increase absolute LV volumes while dobutamine decreased absolute LV volumes. Furthermore, as an estimate of the impact of drug infusion on preload-responsiveness, we also calculated pulse pressure variation during the three conditions during 10 ml/kg tidal volume ventilation. Pulse pressure variation was defined as the ratio of the difference between maximal and minimal arterial pulse pressure averaged over three breaths times 100. Pulse pressure variation for baseline, esmolol, and dobutamine were 11.5 ± 3.5%, 8.0 ± 3.2% and 13.6 ± 6.2%, respectively (NS).

Fig. 4
figure 4

LV pressure loop change induced by inferior vena caval (IVC) occlusion (left) and 5 mmHg CPAP (right) under baseline (control; solid line), esmolol infusion (light dashed line), and dobutamine infusion (heavy dashed line) for the same animal as depicted in Fig. 1. Diagonal lines End-systolic pressure-volume domains for each state

Fig. 5
figure 5

Ees measured at baseline and during esmolol and dobutamine infusions as estimated by in inferior vena caval (IVC) occlusion and continuous positive airway pressure (CPAP)


This study demonstrates that simple inspiratory hold maneuvers can be used to rapidly alter LV filling, allowing the construction of LV ESPVR. However, the relationship between CPAP-derived Ees measures and IVC occlusion-derived measures is both variable and inconsistent. Importantly, in this population of nonvolume expanded subjects the level of CPAP needed to create Ees data does not need be great, thus minimizing the potential risk that such inspiratory hold maneuvers may entail to collect these data. Measures of contractility are often interrelated because of shared common variables. For the same cardiac output LV dilation often occurs in heart failure, whereas LV volumes decrease with increasing contractility. In our study esmolol universally increased LV volume whereas dobutamine decreased them as compared to control, although the Ees values did not consistently vary in a fashion predicted by changing inotropy. Furthermore, the resultant Ees values created by inspiratory hold maneuvers were not the same as LV ESPVR created by IVC occlusions. Still, all Ees values change in a direction similar to changes in contractility during by esmolol and dobutamine. The reasons why Ees values differ between IVC occlusion and CPAP are not clear but probably reflect a combination of altered ITP, ventricular interdependence and afterload created by lung inflation not seen with IVC occlusion. Interestingly, Haney et al. [5] showed an excellent relationship between IVC occlusion-derived and CPAP-derived Ees values as contractility was varied by depth of anesthesia. Presumably, inhalational anesthesia induced a marked decrease in sympathetic tone inhibition, making animals more stable than pentobarbital anesthesia.

Baan et al. [16] and Kass et al. [17] demonstrated that although many preload and afterload varying maneuvers can use used to generate LV ESPVR, the Ees values differ slightly if the manipulations compared alter preload, afterload, or both. However, we found that under both IVC occlusion and CPAP conditions the changes in Ees occur in a predictable fashion to changes in contractility. We consistently saw CPAP increasing LV pressures independent of LV volume (Fig. 1) reflecting unaccounted for associated increase in ITP. Importantly, we calculated CPAP-derived Ees from the steady state decrease in the ESPVR during CPAP and did not include the baseline non-CPAP end-systolic pressure volume point. Since ESPVR is usually calculated by the best-fit slope of this line, excluding non-CPAP data simplifies the subsequent analysis. Increases in ITP decrease LV ejection pressure decreasing LV afterload [18]. However, lung inflation can also change pulmonary vascular resistance, which itself may alter both right ventricular ejection and LV diastolic compliance through ventricular interdependence [19]. If ventricular interdependence were operative during CPAP we would have seen a decreased diastolic compliance. We did not. Importantly, our data agree with those of Crottogini et al. [20] who studied the effect of CPAP on the IVC occlusion included changes in ESPVR, and who documented a relative insensitivity of ESPVR to changes in airway pressure. Although LV pressures increased for the same LV volume, these shifts were in a parallel fashion, equal to the parallel shifts in the LV ESPVR and increased as CPAP increased (Fig. 1), consistent with an unaccounted for increase in ITP, not a change in LV myocardial compliance. The magnitude of the parallel shifts in the LV pressure-volume relationships are consistent with the expected increase in ITP induced in supine dogs with normal lung and chest wall compliance [21].

Importantly, Ees values generated by 5 mmHg (approx. 7 cmH2O) CPAP tended to track IVC occlusion Ees values, whereas 15 mmHg (approx. 21 cmH2O) CPAP Ees values tended to overestimate IVC occlusion Ees values (Table 1). These findings are consistent with an increasing afterload effect seen as CPAP is progressively increased altering the purely preload-reducing effect of both IVC occlusion and 5 mmHg CPAP. Finally, PRSW measures by both IVC occlusion and CPAP tended to be similar than Ees values (Figs. 23) further supporting the hypothesis that both IVC occlusion and inspiratory hold techniques can be used to rapidly alter LV preload. Presumably PRSW is less dependent on changes in the end-systolic pressure-volume pairs and more dependent on global ejection. Thus PRSW changes are a more robust metric when changes in ITP artificially alter LV pressures. Unfortunately, calculation of PRSW requires direct measurement of LV pressure, making it less available at the bedside, since it requires direct LV catheterization. Clinically, 5 mmHg CPAP is routinely applied at the bedside even to unstable patients. Thus there is minimal risk of sustained cardiovascular compromise or hyperinflation if these levels of CPAP were used to generate LV ESPVR.

The study has four major limitations. First, we used an intact canine model under general anesthesia. The associated reduction in sympathetic tone may not model accurately the human condition. Van den Berg et al. [7] showed that in volume resuscitated cardiac surgery patients studied in the immediate postoperative period; CPAP of 5–20 cmH2O did not induce transient decreases in venous return. Thus it is not clear whether similar findings of CPAP-induced transient decreases in venous return occur would occur in routine intensive care unit patients who do not have an expanded intravascular volume. Although if airway pressure is increased high enough, for example, as in recruitment maneuvers to 45 cmH2O, cardiac output routinely decreases [22].

Second, we saw no consistent group difference between IVC occlusion and grouped CPAP-induced Ees measures. Some Ees values were higher with IVC occlusion than CPAP and vice versa. Thus the exact mechanisms by which Ees estimates differ between IVC occlusion and CPAP are unclear. However, Ees values commonly vary widely across subjects, making absolute Ees values of less utility in assessing contractility than the changes in Ees in respond to therapy or time, for which the CPAP method was directionally accurate. Importantly, when both Ees values and absolute LV volume measures were included in the analysis of contractility, decreasing Ees and increasing volumes always reflected esmolol-induced decreased contractility and vice versa, for dobutamine.

Third, we did not measure pericardial pressure, and therefore our measures of LV pressures during CPAP were influenced by an unaccounted for increase in pericardial pressure [18]. CPAP-induced increases in lung volume increase ITP [19]. This series increase in right atrial pressure is the presumed mechanism for the subsequent decrease in LV pressures and volume [17]. We have previously shown that the immediate effect of increasing ITP by a Valsalva maneuver is to increase LV ejection pressure relative to atmosphere, but not alter LV transmural pressure. The greater the degree of ITP increase, the greater is the increase in LV pressure. This is what we observed, as illustrated in Fig. 1. Note that with increasing CPAP the subsequent LV pressure-volume loops are shifted upwards and to the left, with volume on the x-axis. Since Ees is defined by slope and PRSW by pressure-volume loop change, surrounding pressure influences neither of which.

Fourth, in volume-resuscitated patients following cardiac surgery increasing airway pressure by as much as 20 cmH2O did not alter venous return [7]. Thus for the CPAP maneuver to induce acute preload reductions the subjects first need to be preload responsive. Thus if CPAP decreases arterial pulse pressure, some degree of preload-responsiveness should be present. Along the same line of logic, since IVC occlusion induced changes in LV preload must alter LV afterload, LV performance may vary dynamically during the pressure decrease. Since our studies were performed over a very short interval of less than 7 s, changes in autonomic tone were minimal, as supported by the constancy of heart rate during the measurement intervals. Furthermore, Ees values induced by IVC occlusion are the standard method of inducing acute alterations in LV preload in most laboratory and clinical studies [1, 2, 3, 4, 5, 10, 14, 18]. Accordingly, our data suggest that if CPAP induces a decrease in arterial pulse pressure, the resultant estimated ESPVR can be used to estimate LV contractility.


Low levels of CPAP (5 mmHg) may induce enough acute preload-reduction to calculate LV Ees and PRSW in intact dogs. The created Ees values are different but covary with IVC occlusion-derived values. Increasing CPAP to more than 10 mmHg creates estimates of Ees and PRSW that can be inaccurate.