OUP user menu

Impairment of left ventricular function by acute cardiac lymphatic obstruction

L.L Ludwig, E.R Schertel, J.W Pratt, D.E McClure, A.J Ying, C.F Heck, P.D Myerowitz
DOI: http://dx.doi.org/10.1016/S0008-6363(96)00177-0 164-171 First published online: 1 January 1997


Objectives: We performed the following study to define the effects of acute cardiac lymphatic obstruction on left ventricular (LV) systolic and diastolic function. Methods: Cardiac lymphatic obstruction was created in 8 pentobarbital-anesthetized dogs by identifying (Evans blue) and ligating the right and left epicardial lymphatics, the afferent and efferent lymphatics associated with the pretrachael and cardiac lymph nodes, and the thoracic duct. Left ventricular function was assessed by analysis of micromanometer-conductance catheter-derived LV pressure-volume relationships. Contractility was assessed by preload recruitable stroke work (PRSW). The active and passive phases of LV relaxation were assessed by the time constant of isovolumic relaxation (τ) and the end-diastolic pressure-volume relationship (stiffness), respectively. Results: PRSW decreased significantly and τ increased significantly from baseline at 1, 2, and 3 h after cardiac lymphatic obstruction (n = 8), but stiffness did not change. Cardiac lymphatic obstruction had similar effects on LV function in a group of autonomically blocked dogs (n = 5). Left ventricular function did not change in sham treated controls (n = 8). Cardiac lymphatic obstruction induced a significant increase in LV wet/dry weight ratios (3.58 ± 0.01) when compared to the control group (3.53 ± 0.02). Histophatology of the myocardium in the lymphatic obstruction groups revealed significant lymphangiectasis and increased interstitial spacing when compared to controls. Conclusions: Acute cardiac lymphatic obstruction depresses contractility and active relaxation and causes mild LV myocardial edema, but does not alter diastolic stiffness.

  • Ventricular function
  • Lymph flow
  • Edema
  • Dog
  • anesthetized

1. Introduction

Interstitial myocardial edema occurs in numerous clinical conditions and is commonly associated with ventricular dysfunction. However, only recently has adequate evidence been obtained to support the concept of a causal relationship between interstitial myocardial edema and left ventricular dysfunction. In those studies, lowered colloid osmotic pressure [1–3], continuous warm blood cardioplegia[4], and increased coronary microvascular pressure[5, 6] were each found to induce predominantly interstitial myocardial edema and to depress left ventricular contractility. Diastolic function was also found to be depressed by these methods of creating interstitial edema [2, 3, 6]. The demonstration by these studies that different methods of inducing interstitial myocardial edema produce similar changes in ventricular function strengthens the concept of a causal relationship between interstitial myocardial edema and ventricular dysfunction. In this context, acute cardiac lymphatic obstruction has been found to produce histopathologic evidence of interstitial edema [7], but its effects on ventricular function and myocardial water content, as measured by gravimetric means, have not been investigated. Chronic obstruction of cardiac lymph flow results in histopathologic changes in the myocardium and atrioventricular valves consistent with interstitial edema[8, 9], electrocardiographic changes [10, 11], and decreases in developed pressure and the rate of change of pressure[12, 13]. However, detailed examination of the effects of chronic lymphatic obstruction on contractility and diastolic function have not been performed. Cardiac lymphatic obstruction induces interstitial edema by a mechanism that is distinctly different from methods previously utilized and, therefore, establishing its effects on ventricular function may further validate the concept of a causal relationship between interstitial myocardial edema and ventricular dysfunction.

The purpose of our study was to test the hypothesis that acute cardiac lymphatic obstruction induces interstitial myocardial edema and ventricular dysfunction, specifically causing a decrease in contractility and increases in diastolic stiffness and the time constant of isovolumic relaxation (τ). We utilized the load-independent index, preload recruitable stroke work (PRSW), to assess contractility, τ to evaluate active relaxation, and the slope of the end-diastolic pressure volume relationship to assess diastolic stiffness. We controlled for the influences of time, experimental preparation, and the autonomic nervous system on our results by utilizing appropriate control groups.

2. Methods

2.1. Experimental preparation

Twenty-one mature male, heartworm-free, semi-conditioned dogs (weight range 20.5–25.5 kg) were anesthetized by an intravenous dose of sodium pentobarbital (25–35 mg · kg−1) followed by continuous administration (5 mg · kg−1 · h−1). The investigation conformed with the Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health (NIH publication No. 85–23, revised 1985). The dogs were intubated and ventilated with 100% O2 using a volume-cycled ventilator (Model 613, Harvard Apparatus Co., Inc., Millis, MA) at a tidal volume of 15 to 20 ml · kg−1 and a respiratory rate of 8–15 breaths · min−1. The dogs were paralyzed by intravenous administration of pipecuronium (70 μg · kg−1), followed by additional hourly doses (35 μg · kg−1). Maintenance fluids were given as continuous infusion of 0.9% NaCl at a rate of 10 ml · kg−1 · h−1. Arterial pH, Pco2, and PO2 were measured periodically (ABL 30, Radiometer, Denmark) during instrumentation and maintained within normal limits by adjusting tidal volume and ventilatory rate, and by intravenous administration of sodium bicarbonate. The left femoral artery was cannulated for sampling arterial blood. The left femoral vein was cannulated for drug and maintenance fluid administration. A tracheotomy was then performed and a tracheal cannula was inserted and secured. The chest was opened by a median sternotomy and the internal thoracic vessels were ligated and divided when necessary. The heart was cradled in the incised pericardium. Cardiac lymphatic channels were identified by injections of approximately 0.1 ml of Evans blue (0.5% aqueous solution) into the epicardium of the right and left ventricles. Once visualized, the right and left epicardial lymphatics, afferent and efferent lymphatics of the pretracheal and cardiac lymph nodes, and the thoracic duct were isolated and encircled with 5–0 ligatures. A 7.5-F thermodilution catheter was introduced via the right femoral vein and advanced into the pulmonary artery for measuring pulmonary artery pressure. Sodium heparin was administered as an intravenous bolus of 400 U followed by 200 U · h−1. An elastic vessel loop was positioned around the caudal thoracic vena cava as a means of altering loading conditions in the LV.

2.2. Left ventricular function and hemodynamic parameter assessment

Left ventricular contractility was evaluated utilizing preload recruitable stroke work (PRSW) derived from the stroke work-end diastolic volume relationship (SWEDVR)[14]. Left ventricular volume was assessed by determining the electrical conductance of the time-varying quantity of blood in the LV using a 7-F, 12-electrode conductance catheter (Webster Laboratories, Baldwin Park, CA) and stimulator/signal processor (Sigma-5-DF, Leycom, The Netherlands) [15, 16]. The conductance catheter was introduced through the left carotid artery, inserted in the LV and positioned along the long axis of the ventricle such that a maximum number of electrodes were within the LV. Position was verified by evaluating the relationship of segmental volume signals to the electrocardiogram. A 7-F micromanometer catheter (Millar Instruments, Houston, TX) was advanced through the right femoral artery and positioned in the LV for pressure determination. The position of the micromanometer was verified by pressure tracings.

The outermost of the 12 electrodes of the conductance catheter were used to apply 2 opposite-polarity, low-amplitude, high-frequency electrical fields. The remaining electrodes positioned within the LV measured interelectrode segmental voltage differences. Segmental conductances (Gi) were calculated from the segmental voltage differences. The volume of each segment was calculated as: Embedded Image where subscript i refers to segment i; Vi(t), the time varying (intraventricular) segmental volume; ai, the dimensionless slope factor which is assumed to be 1.0; Li, the interelectrode distance; sb, the specific conductivity of blood; Gi(t), the time varying segmental conductance; and GPi, the parallel conductance. These segmental conductance signals were calibrated into time-varying segmental volumes and the total LV volume was calculated as the sum of the 5 segmental volumes. The specific conductivity of a known volume of blood (sb) was determined by withdrawing 4 ml of arterial blood directly into a blood conductivity test chamber. The parallel conductance of structures surrounding the blood in the LV (GPi) was accounted for by transiently changing blood conductivity in the LV by injecting 3 ml of 7% NaCl into the pulmonary artery. Cardiac output was measured by the thermodilution technique using a cardiac output computer (SAT-1, American Edwards Laboratories, Santa Ana, CA). Stroke volume of the conductance catheter system was calibrated utilizing the average value calculated from 3–5 thermodilution measurements of cardiac output and heart rate determined from the LV pressure trace. These latter procedures constituted catheter calibration and were performed prior to each assessment of cardiac function. A complete description of the principles and use of the conductance catheter technique appears elsewhere [15, 17].

Left ventricular function and hemodynamic data were recorded at end-expiration with the ventilator turned off (≤ 15 s). The left ventricular pressure, electrocardiogram and segmental volume signals were digitized at 200 Hz (CONDUCT-PC software, Leycom, The Netherlands) on a 486, 50 MHz microcomputer and stored on hard disc for later analysis. The SWEDVR was obtained by transiently occluding venous return for approximately 10 s and analyzing 8–15 consecutive beats, starting from the point of reduction of LV pressure. A minimum of 3 venous occlusions were performed at each experimental period. Extrasystolic beats and beats where end-systolic pressure decreased below 70 mmHg were excluded from analysis. Stroke work (SW) was calculated as the integral of LV transmural pressure and volume over each cardiac cycle from: Embedded Image PRSW was determined from the slope of the linear regression analysis of the SWEDVR: Embedded Image where Sprsw is the slope of the linear SWEDVR, Ved is the end-diastolic volume and Vosw is the volume axis intercept. The calculated PRSW values with correlation coefficients greater than 0.75 were averaged for each experimental period.

The time constant of isovolumic relaxation was derived from the following relationship by plotting left ventricular dP/dt against pressure [18]. Embedded Image The least square linear regression of the dP/dt and pressure data for the isovolumic period of relaxation yielded a slope (A) equal to - 1/τ and a P-axis intercept of Pasym. The value for τ reported for each experimental period represents an average of the values calculated from each cardiac cycle over a 1 min recording.

Left ventricular compliance was assessed by the slope of the linear regression analysis of the end-diastolic pressure-volume relationship (EDPVR) for the same beats as the SWEDVR such that: Embedded Image where Ped is the end-diastolic pressure, Ved is the end-diastolic volume, and Vod is the volume intercept.

The amplified analog signals of the LV, aortic, pulmonary arterial and right atrial pressure transducers were digitized and acquired by a 16-channel data acquisition-analysis system (PO-NE-MAH, Inc., CT) at 250 Hz for each channel. In addition, the moving average of all derived data was recorded every 5 s for the duration of each experiment. Left ventricular end-diastolic pressure was determined using the LV pressure computer algorithm (PO-NE-MAH, Inc., CT). Mean pressures were derived for the aortic (MAP), pulmonary arterial (PPA) and right atrial (PRA) pressures as calculated over a 1 min recording period.

2.3. Protocol

All dogs were allowed to stabilize for 30 min prior to each experiment and assigned to one of 3 groups: lymphatic obstruction (LO; n = 8), lymphatic obstruction with autonomic nervous system blockade (LO/AX; n = 5), and a control group (n = 8). Autonomic nervous system blockade was performed 20 min prior to baseline measurements by intravenous administration of atropine sulfate (0.2 mg · kg−1 then 0.1 mg · kg−1 · h−1) and propranolol HCl (0.5 mg · kg−1 then 0.25 mg · kg−1 · h−1). Baseline LV function, hemodynamic parameter and blood gas measurements were performed in all dogs at 0 and 30 min. After baseline measurements, lymphatic obstruction was accomplished in the LO and LO/AX groups by tightening the ligatures encircling the lymphatic vessels. Lymphatic obstruction was confirmed with injections of Evans blue after ligation. Left ventricular function, hemodynamic parameter and blood gas measurements were performed at 1, 2 and 3 h. In the control group, left ventricular function, hemodynamic parameter and blood gas measurements were performed as in the other groups, except that the ligatures placed during the preparation period were removed. After all data were recorded, 7 ml of saturated KCl was injected into the peripheral catheter and the heart was rapidly excised. Samples of the LV were taken for wet/dry weight ratio analysis and histophatology.

2.4. Wet/dry weight ratio analysis

The LV samples taken at the end of each experiment were immediately weighed (wet weight). The samples were then scored with a scalpel, to increase surface area for drying, and placed in a 70°C oven where they were kept until a constant weight was reached (dry weight). The wet/dry weight ratio was determined as wet minus dry weight divided by dry weight.

2.5. Light microscopy

Full-thickness tissue samples were obtained from the LV of animals in the control and LO groups for light-microscopic histologic examination. The samples were collected and fixed in 10% buffered formalin before they were sectioned, mounted, and stained with hematoxylin and eosin. All slides were evaluated in a blinded fashion. Each of 5 histopathologic indices was graded on the basis of a scale of normal, mild, moderate, and severe. A numerical grade of 0 was given if the lesion was not present (normal), 1 if the lesion was present and of mild severity, 2 if the lesion was of moderate severity, and 3 if the lesion was of marked severity. The histopathologic indices that were semi-quantitatively evaluated were perivascular interstitial spacing, perivascular and interstitial hemorrhage, lymphangiectasis, inflammation, and fibrin deposition. Endomyocardial and epimyocardial regions were evaluated and the highest score used for analysis.

2.6. Statistical analysis

Data are presented as the mean and the standard error of the mean. Comparisons within and between groups were made using a two-way repeated measures analysis of variance. Student-Neumann-Keuls multiple comparison tests were performed when significant variation occurred within or between groups (P < 0.05). Wet/dry weight ratios were compared between groups with t-tests. Histopathologic index scores of the LO and control group hearts were compared using a Mann-Whitney Rank Sum test.

3. Results

3.1. Hemodynamic parameters (Table 1)

The baseline heart rate of the LO/AX group was significantly less than that of the LO group, but did not differ from that of the control group. Heart rate decreased significantly after lymphatic obstruction in the LO group, but remained stable thereafter. Heart rate did not change in the control or LO/AX groups over the course of the study. Mean arterial pressure did not change during the study in any of the groups and did not differ between groups. Cardiac output declined similarly in all groups over the course of the study, but only significantly in the control and LO groups at 3 h. There were no differences in cardiac output between groups. Pulmonary artery pressure increased significantly from baseline in the control group at 3 h and in the LO group at 2 and 3 h after lymphatic obstruction, but there were no significant differences between groups. Right atrial pressure did not change during the study nor were there differences between groups.

View this table:
Table 1

Heart rate (HR), pulmonary artery pressure (Ppa), right atrial pressure (Pra), mean arterial pressure (MAP) and cardiac output (CO) responses to cardiac lymphatic obstruction in control, lymphatic obstruction (LO), and lymphatic obstruction/autonomically blocked (LO/AX) groups

ParameterGroupBaseline1 h2 h3 hr
HR (beats·min−1)Control143±10143±9147±9151±7
Ppa (mmHg)Control11.5±0.412.4±0.612.8±0.612.6±0.6*
Pra (mmHg)Control0.55±0.290.56±0.250.46±0.310.56±0.42
MAP (mmHg)Control125±9125±10127±9124±9
CO (1·min−1)Control2.47±0.162.27±0.192.2±0.191.87±0.13*
  • * P < 0.05 for difference from baseline period.

    P < 0.05 for difference from control group.

    P < 0.05 for difference between LO and LO/AX groups.

3.2. Contractile function (Fig. GR1)

Preload recruitable stroke work values decreased significantly at 1, 2 and 3 h after lymphatic obstruction in both the LO and LO/AX groups. Preload recruitable stroke work did not change over the course of the study in the control group. There were no significant differences in PRSW between groups. The VOSW of the SWEDVR did not differ between groups or change during the study.

Fig. GR1

Preload recruitable stroke work (PRSW), volume axis intercept of the stroke work-end-diastolic volume relationship (VOSW) and time constant of isovolumic pressure decline (t) responses to cardiac lymphatic obstruction in control (open square), lymphatic obstruction (closed circle), and lymphatic obstruction/autonomically blocked (closed triangle) groups. *P < 0.05 for difference from baseline period; P < 0.05 for difference from control group P < 0.05 for difference between LO and LO/AX groups.

3.3. Isovolumic relaxation (Fig. GR1)

The τ-values increased significantly from baseline after 3 h of lymphatic obstruction in the LO group and 2 and 3 h in the LO/AX group. The τ-values did not change in the control group during the study.

3.4. End-diastolic pressure-volume relationship (Table 2)

There were no significant within or between group differences in LV Ped or Ved during the study. Stiffness, the slope of the EDPVR, did not change in response to lymphatic obstruction.

View this table:
Table 2

Left ventricular end-diastolic pressure (Ped), end-diastolic volume (Ved), and stiffness responses to cardiac lymphatic obstruction in control, lymphatic obstruction (LO), and lymphatic obstruction/autonomically blocked (LO/AX) groups

ParameterGroupBaseline1 h2 h3 h
Ped (mmHg)Control4.5±0.54.5±0.614.1±0.53.3±0.7
Ved (ml)Control35±546±541±1137±5
Stiffness (mmHg·ml−1)Control0.2±0.020.17±0.020.19±0.020.26±0.03
  • * P < 0.05 for difference from baseline period.

    P < 0.05 for difference from control group.

    P < 0.05 for difference between LO and LO/AX groups.

3.5. Wet/dry weight ratios

The wet/dry weight ratios of the LV specimens in the LO (3.58 ± 0.01) and LO/AX (3.58 ± 0.01) groups were significantly greater than those of the control group (3.53 ± 0.02).

3.6. Histopathology

The median scores of the interstitial spacing and lymphangiectasis were 2.0 and 3.0, respectively, in the LO group. These values were significantly greater than their respective values in the control group (1.0 and 1.0). The median scores for perivascular hemorrhage (2.0), inflammation (2.0), and fibrin deposition (0) in the LO group did not differ significantly from their respective control group values (1.5, 1.5, and 0.0). The histopathologic changes of the control animals were found only in the epicardium and were likely due to the handling necessary for surgical preparation.

3.7. Blood gas/acid base values

The Pao2 values did not differ between groups and the average values were above 270 torr in all groups throughout the study. The pH values did not differ between or within groups, and the average values ranged between 7.37 and 7.45 over the course of the study. The Paco2 values did not differ between or within groups, and the average values ranged between 29 and 36 torr over the course of the study.

4. Discussion

Cardiac lymphatic obstruction (CLO) resulted in a progressive decrease in left ventricular contractility and prolongation of isovolumic relaxation, reflecting impairment of the active processes of contraction and relaxation. However, CLO did not significantly affect diastolic stiffness. Cardiac lymphatic obstruction had little influence on the hemodynamic parameters measured. The slight, but progressive decline in cardiac output and increase in pulmonary artery pressure occurred in all groups and was therefore not likely to be the result of CLO or the cause of the function changes in the CLO groups.

Autonomic nervous system mechanisms were not responsible for the changes in systolic and diastolic function, since CLO had similar effects in dogs with β-adrenergic and cholinergic receptor blockade. Left ventricular systolic and diastolic function remained unchanged in dogs of the control group, indicating that the experimental preparation had no effect on these parameters. We did not perform experiments to control for the effects of autonomic nervous system blockade in this study, but have previously performed such experiments using the same autonomic blockade protocol [19]. In that previous study, neither ventricular function nor hemodynamic parameters changed over time in the autonomically blocked control group. Further, autonomic blockade affected heart rate and τ in a manner similar to what was observed in the current study.

While the effects of acute CLO on ventricular function have not been previously reported, there are several reports of the effects of chronic CLO on ventricular function. Ullal and colleagues [13] and Symbas and co-workers [12] have demonstrated in dogs that CLO for 1–30 weeks resulted in a decrease in cardiac output and dP/dt. These studies, however, did not control for autonomic reflexes or evaluate the effects of CLO on diastolic function. Furthermore, the indices used to assess systolic function were load-dependent and limited interpretation of the results. Thus, our study extends previous findings by defining the effects of acute CLO on the major determinants of systolic and diastolic function—contractility, the rate of isovolumic relaxation, and diastolic stiffness.

Acute cardiac lymphatic obstruction produced gravimetric and histopathologic evidence of myocardial edema when compared to a control group. These findings are consistent with the histopathologic evidence of previous studies [7, 12, 13]. However, there have been no previous reports of gravimetric measurement of CLO-induced changes in myocardial water content. The increase in wet/dry weight ratio brought about by CLO (1.4%) was less than what we observed previously (6%) when interstitial myocardial edema was created in dogs by acute coronary venous hypertension [6]. Despite the differing degrees of edema produced by CLO and coronary venous hypertension, contractility decreased and τ increased by similar amounts. A major difference between the results of the two studies was that CLO did not cause an increase in diastolic stiffness. These disparate responses may merely reflect a difference in the degree of edema. Assuming that the edema induced by CLO and coronary venous hypertension was responsible for the LV dysfunction, contractility and τ may be more sensitive to water content changes than stiffness. Thus, the minor increase in water content with CLO may not have been adequate to increase stiffness. Alternatively, CLO may alter ventricular function by a mechanism that differs from that of coronary venous hypertension.

Cardiac lymphatic obstruction may induce LV dysfunction through several possible mechanisms. For the sake of discussion the mechanisms may be divided into those that are and those that are not related to edema. A mechanism which may cause dysfunction that is not necessarily related to edema formation is: acute lymphatic obstruction may cause an increase in interstitial concentration of a metabolite(s) that has negative inotropic effects, but is normally cleared by lymph. The exact nature of the substance can only be speculated, but accumulation of such a substance may occur without a significant increase in water content.

There are several previously proposed mechanisms by which the edema induced by CLO may have caused ventricular dysfunction. Rusznyak [20] and other investigators[10, 21] reported significant electrocardiographic (ECG) changes after CLO that included ST segment changes and alterations in T-wave morphology consistent with “hypoxaemia”. Alternative methods of inducing interstitial myocardial edema, such as coronary sinus occlusion, have also caused ECG changes that included ST segment elevation and altered T-wave morphology [22]. These conduction disturbances may explain the deterioration in systolic function and perhaps active relaxation, but do not as clearly explain edema-related changes in stiffness. We did not perform a detailed analysis of the ECG in our study, but did note occasional premature ventricular contractions. These arrhythmias were seen in dogs of all groups with approximately equal frequency and may have been the result of the catheters in the LV.

A second mechanism by which CLO-induced interstitial myocardial edema may have caused LV dysfunction is by causing oxygen supply limitation and ischemia. Myocardial edema may result in increased coronary vascular resistance and oxygen diffusion distance that limit oxygen delivery and result in ischemia. Recently, Rubboli and colleagues [1] demonstrated that myocardial edema, induced by altering coronary perfusion pressure in the isolated rat heart, resulted in an increase in coronary vascular resistance, a decrease in coronary blood flow, and a decrease in developed pressure. However, the increase in water content in that study was markedly greater than what we observed with CLO in the current study or with coronary venous hypertension [6]. Acute CLO has been demonstrated by Pick and co-workers [23] to create histopathologic evidence of ischemic injury, as assessed by a hematoxylin-basic fuchsin-picric acid stain. Other investigators have demonstrated that obstruction of cardiac lymphatics results in coronary microvasculature injury [24]. Furthermore, in an experiment in which the coronary sinus was occluded in combination with lymphatic obstruction in dogs, there were ECG changes consistent with myocardial infarction, evidence of cardiac muscle necrosis, and an increase in mortality over dogs with either coronary sinus occlusion or CLO alone [25]. Thus, CLO may induce adequate edema over time to cause the changes described above and ischemic injury. In a contrasting study carried out in vivo in dogs using 31P-nuclear magnetic resonance spectroscopy [26], we found that acute coronary venous hypertension and the associated interstitial myocardial edema did not diminish the LV creatine phosphate/adenosine triphosphate concentration ratio. The results suggest that no significant oxygen supply limitation was created by either the altered coronary perfusion pressure or the interstitial edema produced by coronary venous hypertension. Since the increase in myocardial water content in our previous study was approximately 4 times that observed in the present study, it is doubtful that oxygen supply limitation played an important role in the CLO-induced LV dysfunction.

Interstitial myocardial edema also may interfere with systolic and diastolic function by altering the viscoelastic properties of the myocardium. This mechanism may include disruption or acute remodelling of the myocardial extracellular protein matrix [27]. Several studies have demonstrated that interstitial myocardial edema causes an increase in diastolic stiffness [3, 6]. However, little direct evidence is available to support involvement of this mechanism in the decline of systolic function.

In conclusion, acute cardiac lymphatic obstruction resulted in a decrease in contractility and prolongation of active relaxation, but did not affect diastolic stiffness. These changes were associated with the formation of interstitial myocardial edema as demonstrated by gravimetric and histopathologic means. The autonomic nervous system was not involved in the effects of lymphatic obstruction. The role of interstitial edema as a mechanism of the ventricular dysfunction induced by lymphatic obstruction was not clarified by this study. However, this study adds to the evidence supporting a causal association between interstitial myocardial edema and ventricular dysfunction.


We thank Angela Phillips, Kim Watson, and Brian Mitchell for their technical assistance. We appreciate the help of Dr. Steven Weisbrode in performing the histopathology. This study was supported by the Department of Surgery Medical Research Development Fund and Surgical Research Incorporated, Columbus, Ohio.


  • * Corresponding author. N-816 Doan Hall, Department of Surgery, 410 West 10th Avenue, Columbus, OH 43210, USA. Tel. + 1 614 293-4558; Fax + 1 614 293-4726.


View Abstract