Author + information
- Received November 9, 2002
- Revision received January 11, 2003
- Accepted February 20, 2003
- Published online August 20, 2003.
- Stéphane Arquès, MD*,
- Pierre Ambrosi, MD, PhD†,* (, )
- Richard Gélisse, MD*,
- Roger Luccioni, MD, FACC† and
- Gilbert Habib, MD, FACC†
- ↵*Reprint requests and correspondence:
Dr. Pierre Ambrosi, Department of Cardiology, La Timone Hospital, Boulevard Jean Moulin, 13385 Marseille, France.
Objectives This study evaluated the relative contribution of serum colloid osmotic pressure (COP) lowering and pulmonary artery wedge pressure (PAWP) elevation in the pathogenesis of pulmonary edema in patients with systolic or isolated diastolic heart failure (DHF).
Background The role of hypoalbuminemia and the resulting low COP have been shown in some patients with acute systolic heart failure (SHF).
Methods Colloid osmotic pressure and PAWP were determined in 100 patients with acute heart failure (HF) (56 with DHF and 44 with SHF; mean age, 78 ± 12 years), in 35 patients with acute dyspnea from pulmonary origin, and in 15 normal controls. Pulmonary artery wedge pressure was estimated using transthoracic Doppler echocardiography.
Results Colloid osmotic pressure was significantly lower in the DHF group (20.5 ± 5 mm Hg) than in the SHF group (24.2 ± 3.7 mm Hg, p < 0.001), pulmonary disease group (25.1 ± 4.2 mm Hg, p < 0.001), or normal control group (24.7 ± 3 mm Hg). Low COP resulted from hypoalbuminemia due to age, malnutrition, and sepsis. Pulmonary artery wedge pressure was significantly higher in patients with SHF (26 ± 6.3 mm Hg) than in the patients with DHF (20.3 ± 7 mm Hg, p < 0.001) and was significantly higher in the patients with DHF than in the patients with pulmonary disease (13 ± 4.2 mm Hg, p < 0.001). The COP–PAWP gradient was similar in patients with SHF (−1.6 ± 7.1 mm Hg) and patients with DHF (0.7 ± 6 mm Hg).
Conclusions Frequent hypoalbuminemia resulting in low COP facilitates the onset of pulmonary edema in patients with DHF who usually have lower PAWP than patients with SHF.
According to Starling's hypothesis, the main cause of pulmonary edema is a decrease in the serum colloid osmotic pressure–pulmonary artery wedge pressure (COP–PAWP) gradient (1). Critical decrease in the COP-PAWP gradient is usually related to a major increase in PAWP due to left ventricular (LV) failure. In 1978, Weil et al. (2), then in 1982, Rakow et al. (3)demonstrated that in some cases a substantial decline in COP can be an additional cause for the decrease of this gradient and favors acute exacerbation of chronic heart failure (HF) (2,3). In these early works, diastolic heart failure (DHF) was not clearly identified as a cause of frequent pulmonary edema, and the relative contributions of COP and PAWP abnormalities in the pathogenesis of acute DHF have not been evaluated. Despite the difficulties in affirming the diagnosis of DHF, there is now much evidence that in nearly half of the patients presenting with pulmonary edema, HF is due to isolated diastolic dysfunction (4).
According to experimental and clinical studies, COP may be accurately estimated from albuminemia and protidemia (5,6). More recently, Doppler echocardiography has been validated for the estimation of PAWP, using new combined indexes that can be used in the emergency care unit (7,8). The aim of this study was, therefore, to assess the relative importance of COP and PAWP in acute DHF compared with acute systolic heart failure (SHF), using serum protein concentration and Doppler echocardiography to estimate the imbalance of Starling forces.
Patients with acute HF
We included 100 consecutive patients admitted for acute dyspnea with the final diagnosis of acute congestive left-HF, in the absence of severe mitral valve disease, rest angina, or acute myocardial infarction (9).
The diagnosis of acute congestive HF was based on Framingham criteria, the absence of patent primary pulmonary disease, the response to diuretics, and the result of further cardiac imaging (9). All the patients had congestive signs at auscultation and pulmonary edema on chest X-ray. In all the patients, transthoracic Doppler echocardiography was performed within 30 min after the initiation of loop diuretics. The diagnoses of DHF and SHF were based on the presence at echocardiography of LV ejection fractions >50% and <50%, respectively. The presence of malnutrition was assessed according a subjective method (10). Sepsis was defined by the association of fever, C-reactive protein >100 mg/l, and a diagnosis of infection at discharge.
Patients in the pulmonary disease group
Pulmonary artery wedge pressure was measured in 35 patients who suffered from acute dyspnea from various pulmonary causes.
Normal control group
Pulmonary artery wedge pressure was also measured in 15 elderly subjects (8 women, 7 men) without overt cardiovascular or pulmonary disease.
Total serum protein (P) and serum albumin (A) concentrations were obtained from venous blood sampled at the admission in all the patients. Serum protein was measured using the Biuret method (Vitros; normal value between 6 and 8 g/dl), and A was measured using the immunoturbidimetry method (Behring turbidimeter; normal value between 3.5 and 5.0 g/dl). Serum globulin concentration (G) was calculated as P − A. We used the Landis-Pappenheimer formula to estimate COP: (mm Hg) = A/P × (2.8 × P + 0.18 × P2+ 0.012 × P3) + G/P × (1.6 × P + 0.15 × P2+ 0.006 × P3), with A, G, and P in g/dl. A previously reported normal value of COP was 25.4 ± 2.3 mm Hg (5). Creatinine clearance was determined using the Cockcroft formula. Early variation of hematocrit (ht) was measured between the admission and 72 h (or 48 h when ht at 72 h was not available).
All Doppler echocardiography examinations were performed by the same operator (S.A.) using either ESAOTE Challenge (ESAOTE Biomedica, Italy) or ALOKA SSD5500 PHD (Aloka Co. Ltd., Tokyo, Japan) ultrasound systems with 2.5 MHz and 3.5 MHz transducers. Two-dimensional and Doppler recordings were acquired in apical views. Left ventricular ejection fraction was determined with the Simpson biplane method from two-dimensional images. The isovolumic relaxation time (IRT) (ms) was measured from the end of aortic flow to the onset of mitral flow. The velocity of color M-mode Doppler mitral flow propagation (Vp) (cm/s) was recorded in an apical four-chamber view, positioning the M-mode cursor through the longest column of color flow from the mitral annulus to the apex; Vp was measured as the slope of the first aliasing velocity during early filling, from the mitral valve plane to 4 cm distally into the LV cavity (7,8). The slope of the transition from no color to color was assumed to be Vp when peak velocity (E) was lower than aliasing velocity. Results of three to five consecutive beats were averaged. Pulmonary artery wedge pressure (mm Hg) was estimated using the equation proposed by Gonzalez-Vilchez et al. (8): PAWP = 4.5 × 1,000/(2 × IRT + Vp) − 9. Mitral inflow patterns were defined as: normal pattern by E/A >1 and E/V p < 2; abnormal relaxation pattern by E/A <1; normalized pattern by 1 < E/A < 2 and E/Vp > 2; and restrictive pattern by E/A > 2. Diastolic dysfunction was defined by abnormal relaxation, normalized, or restrictive patterns at Doppler mitral inflow examination (11). Doppler echocardiography recordings were read independently by two echocardiographers. Intra- and inter-observer variability of PAWP was calculated in 15 consecutive patients.
All descriptive data are given as mean ± SD. Inter-group comparison for clinical, biological, and hemodynamic data used one-way analysis of variance. Inter-group comparison for the percentages of hypoalbuminemia and main mechanism of pulmonary edema used a chi-square test. Linear regression analysis was used to assess the relationship between albuminemia and age, C-reactive protein, or creatinine clearance. A p value <0.05 was considered statistically significant.
There were 56 patients in the DHF group (mean age = 79 ± 12 years), 44 in the SHF group (mean age = 76 ± 12 years), 35 in the pulmonary disease group (mean age = 78 ± 11 years), and 15 in the normal control group (75 ± 8 years) (Table 1).
Doppler estimate of PAWP and diastolic function
The feasibility of PAWP measurement was 91% (Table 2). Correlation coefficients were 0.98 for intra-observer and 0.96 for inter-observer comparison of PAWP.
The PAWP was significantly higher in SHF (26 ± 6.3 mm Hg) than in DHF (20.3 ± 7 mm Hg, p < 0.001); PAWP was significantly higher in the DHF group than in the pulmonary disease group (13 ± 4.2 mm Hg, p < 0.001); and PAWP was slightly higher in the pulmonary group than in the normal control group (10.7 ± 2.1 mm Hg, p = 0.044). A generally accepted cut-off value of PAWP for the diagnosis of cardiogenic pulmonary edema in patients presenting with acute dyspnea is ≥18 mm Hg. A PAWP ≥18 mm Hg was found in 98% of the SHF, 60% of the DHF, 12% of the pulmonary disease, and none of the normal control patients (Fig. 1).
Mitral inflow pattern was not obtained in 42 patients, because of atrial fibrillation, cardiac pacing, or E/A fusion (Table 2). Among the 93 other patients, the mitral inflow pattern suggested diastolic dysfunction in 100% of the patients with SHF, 95% of the patients with DHF, and 85% of the patients with pulmonary disease (Table 2). Pseudo-normal and restrictive patterns were significantly more frequent in the SHF and DHF groups than in the pulmonary disease group (p < 0.001).
Albuminemia was significantly lower in the DHF group (2.58 ± 0.7 g/dl) than in either the SHF group (3.2 ± 0.54 g/dl), pulmonary disease group (3.18 ± 0.65 g/dl), or normal control group (3.44 ± 0.43 g/dl) (p < 0.001), with no significant difference between the SHF and pulmonary disease groups (Table 3). Albuminemia <3 g/dl was present in 38 of the 56 patients with DHF (68%), 14 of 44 patients with SHF (32%, p < 0.001 vs. DHF), and 11 of 35 patients with pulmonary disease (31%, p < 0.001 vs. DHF). Main causes of hypoalbuminemia <3 g/dl in the DHF group were malnutrition in 77% and/or sepsis in 41% of the patients. Albuminemia was significantly inversely correlated to age (r = −0.24, p < 0.005) and to plasma C-reactive protein concentration (r = −0.37, p < 0.001). There was no significant correlation between albuminemia and creatinine clearance (p = 0.52); COP was significantly lower in DHF (20.5 ± 5 mm Hg) than in SHF (24.2 ± 3.7 mm Hg, p < 0.001) or pulmonary disease patients (25.1 ± 4.2 mm Hg, p < 0.001), with no significant difference between SHF and pulmonary disease patients (p = 0.29). Previously, a COP ≤18 mm Hg has been shown to facilitate pulmonary edema in critically ill patients (12). A COP ≤18 mm Hg was present in 7% of patients with SHF and 6% of pulmonary disease patients, compared with 34% of the patients with DHF (p < 0.001 vs. SHF and pulmonary disease patients) (Fig. 1). Hematocrit did not show significant variations between admission (ht0) and 72 h (ht72) (SHF: ht0 = 38.9 ± 5.8%, ht72 = 38.5 ± 5.8%; DHF: ht0 = 36.2 ± 5.4%, ht72 = 34 ± 9.2%; pulmonary disease: ht0 = 41.1 ± 5.8%, ht72 = 39.6 ± 5.7%).
Colloid osmotic–hydrostatic pressure gradient
The COP–PAWP gradient was significantly lower in the DHF and SHF than in the pulmonary disease or normal control groups (Table 3). In the previous invasive studies, a value of COP–PAWP gradient ≤6 mm Hg was shown to be associated with pulmonary edema with a high sensitivity and specificity (2,3). A COP–PAWP gradient ≤6 mm Hg was present in 48 patients (86%) with DHF, 40 patients (90%) with SHF, one patient with pulmonary disease, and one control, respectively.
In the present work, we used Doppler echocardiography to measure PAWP in 100 patients at the acute phase of pulmonary edema. This method appeared to be feasible in most of the patients (91%) in this setting. The reliability of Doppler echocardiography for the measurement of PAWP has been assessed by previous studies, by comparison with invasive measurements (7,8,11). For this measurement, we used a validated index derived from Weiss' formula (8,13). The elevation of PAWP correlated well in our study with the abnormality of E/A ratio.
Our first important finding is that PAWP is usually lower in patients with pulmonary edema due to isolated diastolic dysfunction than in patients with pulmonary edema and systolic dysfunction. Moreover, PAWP was higher in patients with DHF than in patients with acute dyspnea of pulmonary origin or in normal controls.
Our study strongly suggests that a major additional mechanism for pulmonary edema in DHF is low oncotic pressure. First, we demonstrated that hypoalbuminemia resulting in low oncotic pressure is significantly more common in patients with acute DHF than in patients with SHF or patients with pulmonary disease. Albuminemia <3.0 g/dl was present in about one-third of the patients with SHF or pulmonary disease and two-thirds of the patients with DHF. Second, as a result of the low oncotic pressure, the gradient of Starling's imbalance forces (oncotic pressure–PAWP) was similarly lowered in the two groups (diastolic and systolic) with pulmonary edema. Thus, this gradient was lower than 6 mm Hg, an accepted threshold value for the setting of pulmonary edema, in most of the patients with DHF (2,3).
In most cases, hypoalbuminemia was related to malnutrition or sepsis and was not the consequence of hemodilution, because diuretic therapy did not increase ht. Consistently, albuminemia was inversely correlated with C-reactive protein and age.
The study population was an unselected sample of patients presenting for acute dyspnea with a mean age very similar to the age of patients hospitalized for acute HF in France or the U.S. (14,15). Thus, our results suggest that hypoalbuminemia is a frequent and major cause of exacerbation of DHF, besides acute elevation of blood pressure, renal failure, tachycardia, ischemia, or high sodium intake. This mechanism may, at least partly, explain why serum albumin concentration is a strong predictor of high mortality in HF (16,17). Through a similar mechanism, hypoalbuminemia has been recently proposed as facilitating acute respiratory distress syndrome (18). To confirm the role of hypoalbuminemia in the exacerbation of DHF, we need further interventional studies evaluating the effects of albumin infusions or of improved nutrition in patients with DHF and severe hypoalbuminemia.
- serum albumin
- serum colloid osmotic pressure
- diastolic heart failure
- serum globulin concentration
- heart failure
- isovolumic relaxation time
- left ventricular
- serum protein
- pulmonary artery wedge pressure
- systolic heart failure
- velocity of color M-mode Doppler mitral flow propagation
- Received November 9, 2002.
- Revision received January 11, 2003.
- Accepted February 20, 2003.
- American College of Cardiology Foundation
- Guyton A.C.,
- Lindsey A.W.
- Banerjee P.,
- Banerjee T.,
- Khand A.,
- Clark A.L.,
- Cleland J.G.
- Morissette M.P.
- Mullins R.E.,
- Pappas A.A.,
- Gadsden R.H.,
- Vence-Pastor D.E.
- Garcia M.J.,
- Ares M.A.,
- Asher C.,
- Rodriguez L.,
- Vandervoort P.,
- Thomas J.D.
- Gonzalez-Vilchez F.,
- Ares M.,
- Ayuela J.,
- Alonso L.
- Remme W.J.,
- Swedberg K.
- Detsky A.S.,
- Baker J.P.,
- Mendelson R.A.,
- Wolman S.L.,
- Wesson D.E.,
- Jeejeebhoy K.N.
- Garcia M.J.,
- Thomas J.D.,
- Klein A.L.
- Senni M.,
- Tribouilloy C.M.,
- Rodeheffer R.J.,
- et al.