Susceptibility of the right and left canine atria to fibrillation in hyperglycemia and hypoglycemia

Journal of Electrocardiology

Vol. 26 No. 2 1993

Susceptibility of the Right and Left Canine Atria to Fibrillation in Hyperglycemia and Hypoglycemia

Panos E. Vardas, MD, Kostas Vemmos, MD, Dimitris A. Sideris, MD, and Spyridon D. Moulopoulos, DD

Abstract: The objective of this study was to investigate the changes in the refractory period and in the susceptibility to fibrillation of canine atria associated with different levels of glycemia, and the differences in these parameters between the two atria. In 20 anesthetized, open-chest dogs weighing 24 kg, the effective refractory period was measured by atria1 pacing with a run of 8 stimuli (S 1-S 1 350 ms) followed by a progressively earlier S2 until no stimulation of the atria1 tissue occurred. The susceptibility to fibrillation was assessed by applying DC at 2, 3, and 4 V for 3 seconds, 7 times each, on the atria1 appendage. If fibrillation occurred and persisted for 3 minutes, a transthoracic synchronized shock was delivered (200 J). The refractory period and the susceptibility to fibrillation were assessed under normoglycemia first, and then under hypo and hyperglycemia, in the right and left atrium successively, in random order. The incidence of induced atria1 fibrillation in the right atrium was: hypoglycemia 3 1.96% ( 132 of 413 attempts); normoglycemia 24.11% (81 of 336; p < 0.05); and hyperglycemia 20.23% (85 of 420). Results for the left atrium were hypoglycemia 52.06% (215 of413); normoglycemia40.18% (135 of336; p < 0.005); andhyperglycemia 32.86% (138 of 420; p < 0.05). Sustained atria1 fibrillation (>3 minutes) occurred significantly more often under hypo rather than hyperglycemia and stimulated the left rather than the right atrium. The refractory period was shortest under hypoglycemia in the left atrium and longest under norm0 or hyperglycemia in the right atrium. The susceptibility to atria1 fibrillation and the propensity to sustained atria1 fibrillation are higher under hypo than hyperglycemia and in the left rather than the right atrium. A shorter refractory period is probably the key factor in these differences. If these findings from healthy dogs apply in humans with paroxysmal atria1 fibrillation, hypoglycemia should probably be considered as a triggering factor. Key words: atria1 fibrillation, glucose, effective refractory period, hypoglycemia, hyperglycemia.

Several factors may affect the susceptibility of the atria to the induction and maintenance of fibrillation. From the Department of Clinical Therapeutics, Athens University, Athens, Greece.

The roles played by these factors or conditions (such as the autonomic nervous system or electrolyte disturbances) have been widely investigated. However, the effect of changes in glycemia levels on the electro-

School of Medicine,

physiology

Reprint requests: Panos E. Vardas, MD, Department of Clinical Therapeutics, Alexandra Hospital, 80 E. Venizelou and K. Lourou Street, GR-115 28, Athens, Greece.

147

of the atria and, in particular,

their sus-

ceptibility to fibrillation remains unknown.

Hypogly-

cemia might

since it is

increase

atria1 susceptibility,

148 Journal of Electrocardiology

Vol. 26 No. 2 April 1993

often accompanied by an increase in serum catecholamine levels and electrolytic disturbances such as hypokalemia and hypomagnesemia. ‘-’ This assumption has been partly confirmed by other observations, which have established that hypoglycemia leads to sinus nodal tachycardia, ventricular ectopic beats, or, rarely, to episodes of atrioventricular nodal tachycardia.6-9 The purpose of this study was to investigate the effect of glycemia levels on the development and persistence of atria1 fibrillation in each atrium and to evaluate any difference between the two atria.

Materials and Methods

ment, using a previously calibrated photometric system. Glucose levels <80 mg/lOO ml were considered hypoglycemic and levels > 120 mg/lOO ml hyperglycemic, while levels between 80 and 120 mg/lOO ml were considered normal. Crystalline insulin (Actrapid Lab, 1 U/kg) was used to induce hypoglycemia, and intravenous dextrose (3 5 % up to 50 ml over 30 minutes) was used to induce hyperglycemia. Serum electrolytes (mixed venous blood; flame photometric method), potassium, and sodium levels were measured before the thoracotomy and each time the glycemia level or the site of stimulation was changed, that is, six times in each typical experiment. The pH, P02, and PCOZ were measured in the arterial blood once for each glycemia level in five experiments only.

Experimental Preparation Twenty mongrel dogs weighing 2 1-3 1 kg were anesthetized with intravenous sodium pentobarbital (25 mg/kg). Additional pentobarbital was given as needed to maintain anesthesia. All dogs had continuous recordings of the electrocardiogram (ECG) (lead II). The recordings were displayed on a recorder (Electronics for Medicine) with a paper speed of 100 mm/s. Intubation and artificial respiration with a Harvard respirator using room air were followed by median thoracotomy and pericardiotomy. The care and use of the animals conformed to the guiding principles of law 1197/8 1 of the Greek national legislation Custom-made metal crocodile electrodes were used for the electrophysiological study and were attached epicardially to the left and right atria1 appendages. Special care was taken that the electrodes would not damage the atria1 tissue. The effective refractory periods of the two atria were measured using a Medtronic 5328 device, while the electrical stimulation used for inducing atria1 fibrillation was supplied as DC from a custom-made rectifier. This latter device had a timer preset for 0.5, 1, 2, 3, and 4 seconds and a voltage output switch set for 1,2, 3, and 4 V. The current supplied (mA) was measured by an ammeter included in the circuit.

Measurements The following biochemical and electrophysiological measurements were made in each experiment. Biochemical. Serum glucose (mixed venous blood) was measured each time the DC intensity was changed, that is, 18 times in each typical experi-

Electrophysiological Study. The heart rate was measured just before the application of DC in all instances. The mean value of the corresponding 21 measurements was taken as the heart rate for each glycemia level and atrium concerned. The effective refractory period and susceptibility of each atrium to fibrillation were measured for each level of glycemia. The measurements were initially made under normal glycemia and subsequently under hypo and hyperglycemia alternatively in successive experiments. The choice of atrium was also random in each case. In cases where the animal was already hypo or hyperglycemic, the measurements were made in the initial state and then in the opposite state, and no measurements were made for normoglycemia. In all experiments, measurements were made only after the glycemic state had been stabilized for at least 40 minutes. Eflective Refractory Period. For the reliable measurement of the effective refractory period, atria1 pacing was applied with 8 beats at a steady rate (S 1-S 1 3 50 ms) in order to depress the sinus rhythm, followed by an early stimulation (S2). The sequence was repeated with S 1-S2 being progressively shortened in steps of 5 ms. The pulses supplied were square waves of 2 ms duration and twice the pacing threshold in intensity. The effective refractory period was taken as the longest S 1-S2 at which the S2 failed to stimulate the atria1 tissue. Stlsceptibility to Fibrillation. In each glycemic state 21 attempts were made on each atrium to induce atria1 fibrillation. These attempts were divided into three groups of seven as follows: 2, 3, and 4 V DC for 3 seconds each time. This type of stimulation was preferred to avoid gross hemodynamic effects from a longer stimulation such as by stimulating the atria at a high rate for 10 seconds. A self-perpetuating arrhythmia with a frequency of >400 beats/mm and

Atrial Fibrillation

lasting for at least 1 second after the offset of the DC was defined as atria1 fibrillation. Atria1 fibrillation lasting more than 30 seconds was considered to be sustained and was terminated by synchronized de% brillation (200 J) applied to the external thoracic wall to avoid damage to the myocardium. The duration of the episodes of sustained or nonsustained atria1 fibrillation was measured in seconds from the time the stimulation ceased.

and Glycemia

l

Vardas et al.

149

mia, 95- 120 mg/lOO ml during normoglycemia, and 260-300 mg/lOO ml during hyperglycemia, with clear gaps between the three glycemic levels. The K+ levels were slightly but significantly higher, the Na+ levels were lower, and the pH levels were higher in normoglycemia than in both hypo and hyperglycemia. The difference in blood gases were not significant between the three glycemic phases.

Current Intensity Statistics Nonparametric changes were analyzed using the chi-square test. For parametric changes the paired or unpaired t test was used. The susceptibility to atria1 fibrillation for each glycemic state was evaluated in two ways: as the incidence of fibrillation out of seven DC applications of the same intensity in each experiment, and as the lowest current intensity needed to cause fibrillation in least in one of the seven applications in each glycemic level and atrium.

The intensities applied during hyperglycemia in both atria (4.77 + 1.90 mA) and to the right atria under all three glycemic conditions (4.73 + 1.82 mA) were slightly higher than during normoglycemia (4.56 k 1.81 mA; p < 0.05) and to the left atria (4.55 + 1.90 mA; p < 0.025), respectively. Other differences in current intensity were not significant.

Heart Rate The heart rate was significantly (p < 0.001) higher in hypoglycemia ( 17 1.8 + 6.6/min) than in normoglycemia ( 156.5 + 9. Urnin), with hyperglycemia (161.7 -t 8.6/min) lying in between.

Results Biochemical Parameters The mean

levels of glucose,

electrolytes

(K+,

Na+ ), and blood gases associated with the three lev-

els of glycemia are shown in Table 1. The glucose levels were fairly steady during each glycemic state, varying from 50 to 80 mg/lOO ml during hypoglyceTable 1.

Mean Levels of Glucose, Electrolytes, and Blood Gases

Parameter

Hypoglycemia

Glucose ” K+ n Nat n PH FO,

64.9 (5.6%)* 118 3.16 (.16%)* 40 145.0 (2.0%)* 40 7.38 (.03%)* 80.6 (7.5%) 5

Normoglycemia 107.2

(9.8%) 96 4.04 (.45%) 32 142.1 (2.2%) 32 7.41 (.02%) 85.5

n PC02 n

5 34.4 (7.1%) 5

35.5

(4.8%) 5 (3.8%) 5

Hyperglycemia 279.6 3.68 143.6 7.39 84.8 31.5

(11.9%)* 120 (.21%)* 40 (2.54%)* 40 (.02%)* 5 (7.6%) 5 (3.7%) 5

Levels of glucose (mg/lOO ml), K+, Na+ (mEq/l), pH, PO*, and PC01 (mmHg) in the three phases of glycemia. Results are mean (SD). * Significant difference from normoglycemia (unpaired t test for glucose and paired t test for other parameters). n = number of measurements.

Effective Refractory Period For each glycemic level the left atrium had a shorter effective refractory period than the right one (Table 2). Also, the effective refractory period for each atrium was shorter during hypoglycemia than either normo or hyperglycemia (the difference between the latter two was not significant).

Susceptibility to Fibrillation At normal glucose levels, that is, 107.2 (9.8) mg/ 100 ml, DC was applied 336 times to the right and left atria (4 of the 20 animals were hypoglycemic at the start of the experiment and so there was no DC application in the normoglycemic state). In hypoglycemia with serum glucose of 64.9 (5.6) mg/lOO ml, DC was applied 413 times to the right and left atria. In one case, the final seven stimulations (4 V/3s) were omitted since ventricular fibrillation occurred at the two lower potential levels. In hyperglycemia with serum glucose of 279.6 (11.9) mg/lOO ml, DC

150

Journal of Electrocardiology

Vol. 26 No. 2 April 1993

Table 2. Effective Refractory Periods Serum Glucose status Hypoglycemia Mean (SD) n P Normoglycemia Mean (SD) n P Hyperglycemia Mean (SD) Alfl Mean (SD) n

Right Atrium

P

Left Atrium

Both Atria

105.3 (13.5%) 20 CO.02


86.3 (6.6%) 20 CO.005

95.8

118.0 (14.2%) 16 NS

CO.01

102.7 (14.4%) 16 NS

110.3

(16.2%) 32 NS

117.8 (7.3%) 20

CO.005

105.0

(11.5%) 20

111.4

(11.6%) 40

113.4 (13.3%) 56

CO.001

97.7

(13.9%) 56

105.6

(15.7%) 112

was applied 420 times to the right and left atria. Atria1 fibrillation occurred 786 times out of a total 2,338 attempts to induce arrhythmia (33.62%). The incidence of fibrillation was significantly higher when DC was applied to the left atrium rather than the right for all levels of glycemia (Table 3). For each atrium the incidence of fibrillation was highest in hypoglycemia and lowest in hyperglycemia-these differences were significant except those between normo and hyperglycemia for the right atrium. Considering all experiments collectively, the incidence of atria1 fibrillation for each glycemic level and site of DC application was a function of the current intensity. The higher the voltage applied the higher the incidence of atria1 fibrillation for both atria (p < 0.01 to p < 0.0005; Table 3). Thus, the relationship between the incidence of atria1 fibrillation as a function of DC intensity was a statistical one rather than an all or none type. However, the lowest DC intensity needed to cause at least one episode of atria1 fibrilla-

tion (out of 7 applications) depended on the glycemic status and the site of the stimulation (Table 4). For any glycemic level, the lowest fibrillogenic DC intensity was higher for the right atrium than for the left one. Similarly for the same atrium, the lowest current needed to cause fibrillation was lowest in hypoglycemia and highest in hyperglycemia, normoglycemia lying in between. The atria1 fibrillation tended to stop spontaneously, although not in every case. Sustained (>30 seconds) fibrillation occurred 90 times (3.85% of all DC applications). The incidence of sustained atria1 fibrillation was higher for the left atrium than for the right one at all levels of glycemia (Table 5). Also for each atrium, the incidence of atria1 fibrillation was highest in hypo and lowest in hyperglycemia (significant differences except between hypo and normoglycemia for the right atrium). The duration of the nonsustained episodes of atria1 fibrillation varied from 1 to 23 seconds. No self-ter-

Table 3. Incidence of Atria1 Fibrillation (%) in Voltage

Atrium

2v

Right % Left % Right % Left % Right % Left % Right %

221140 15.71 461140 32.86 50/140 35.71 821140 58.57 601133 45.11 871133 65.41 1321413 31.96

P Left %

<0.0005 2151413 52.06

3v

4v

All

&i-square

test performed.

Hypoglycemia

P NS <0.02 <0.05 <0.05 NS NS <0.02

<0.005

(14.2%) 40 -Co.001

Each

Voltage,

Normoglycemia 12/l 12 14.29 21/l 12 18.75 271112 24.11 50/l 12 44.64 42/l 12 37.50 641112 57.14 811336 24.11 <0.0005 135/336 40.18

Atrium,

and Glycemic

Level

P

Hyperglycemia

NS

13/140 9.29 32/140 22.86 26/140 18.57 451140 32.14 46/140 32.86 6 l/40 43.57 851420 20.24 <0.0005 138/420 32.86

NS NS <0.05 NS <0.05 NS

<0.05

All 471392 11.99 991392 25.26 103.39 26.28 1771392 45.15 1481385 38.44 2121385 55.06 298/l 169 25.49 <0.0005 488/l 169 41.75

Atrial Fibrillation

and Glycemia

l

Vardas et al.

151

Table 4. Intensity Level Required to Cause Atrial Fibrillation Serum Glucose

Right Atrium

P

Hypoglycemia n P Normoglycemia n P Hyperglycemia

4.14 (1.55%) 20 NS 4.49 ( 1.69%) 16 co.01 5.17 (1.85%)


Al: n

4.61 ;?:73%) 56


co.05

co.02

Left Atrium

Both Atria

2.93 (1.21%) 20 co.01 3.70 (1.59%) 16 NS 3.82 (2.03%)

3.53 (1.50%) 40 co.01 4.09 (1.66%) 32 co.07 4.49 (2.04%) 40 4.04 ( 1.79%) 112

3.47 ;:67%) 56

Mean (SD) intensity (mA) required to cause atria1 fibrillation in each atrium and glycemic level, and significance of mean differences. Paired t test performed. n = number of measurements.

minating episode of atria1 fibrillation lasting over 23 seconds was observed, given that the experimental protocol called for cardioversion after 3 minutes. The longest nonsustained episode (23 seconds) occurred in hypoglycemia with DC application on the left atrium. The probability that atria1 fibrillation, once begun, would be persistent was highest in hypoglycemia and lowest in hyperglycemia (Table 6), although the difference between hypo and normoglycemia was not significant for either atrium. For any given glycemic level, the incidence of sustained atria1 fibrillation was higher for the left than for the right atrium (the difference was significant for hypo and normoglycemia)

.

Effective Refractory Period and Susceptibility to Atrial Fibrillation There was a significant (p < 0.001) negative correlation between the effective refractory period and the incidence of both total atria1 fibrillation episodes and episodes of sustained atria1 fibrillation. Thus, shortening of the effective refractory period by 10 ms increased the incidence of atria1 fibrillation episodes

by approximately 1.26 times and the incidence of sustained atria1 fibrillation by approximately 0.41 times over the total of 21 DC applications. On the other hand, when the effective refractory period was greater than 130 ms there were no episodes of sustained atria1 fibrillation.

Discussion A profibrillatory effect of hypoglycemia on the atria and an antifibrillatory effect of hyperglycemia were evidenced in these experiments by several observations. Hypoglycemia was associated with an increased incidence of atria1 fibrillation compared to hyperglycemia, with normoglycemia lying in between. Furthermore, hypoglycemia was associated with profibrillatory electrophysiologic changes, similar to a shortened refractory period”,” and a lowered current intensity needed to induce atria1 fibrillation, while hyperglycemia had the opposite effect.

Table 6. Probability of Sustained Atrial Fibrillation Serum Glucose Status

Table 5. Incidence of Sustained Atria1 Fibrillation

(%) of Each Atrium and Glycemic Level Serum Glucose Status Hypoglycemia % P Normoglycemia % P Hyperglycemia % All %

Right Atrium

P

10/143
L.&t Atrium 48/413 11.62 co.05 231336 6.84
Both Atria 581826 6.25 CO.025 271672 4.02
Hypoglycemia % time (seconds) P Normoglycemia % time (seconds) P Hyperglycemia % time (seconds) All % time (seconds)

Right Atrium lo/132 7.58 14 NS 4194 4.94 14 co.05 O/85 0 11 14/311 4.50 14

Chi-square test performed.

P
CO.025

NS
Left Atrium

Both Atria

4812 17 22.12 23 NS 231136 16.91 14 <0.0005 51318 3.62 16 76167 1 11.33 23

581349 16.62 23 NS 271230 11.74 16 <0.0005 51403 1.24 16 901982 9.17 23

152

Journal of Electrocardiology

Vol. 26 No. 2 April 1993

The profibrillatory effect was due to hypoglycemia rather than hyperinsulinemia, the latter existing with hyperglycemia also when dextrose followed the insulin administration. The biochemical changes seem also irrelevant since they were either towards the same direction in both hypo and hyperglycemia (K + , Na+, pH) or not significant (PO1, PC02). The increased sinus rate during hypoglycemia might suggest that hypercatecholaminemia, present in hypoglycemia, ’ 2*13 caused atria1 fibrillation via shortening of the effective refractory period. However, during an antifibrillatory hyperglycemia the heart rate was also slightly higher than in normoglycemia. For the same glycemic level, and, presumably, the same blood biochemistry, it was conl%-rned that the effective refractory period of the left atrium is usually shorter than that of the right one as suggested by others.14-l6 Also, the lowest DC intensity that might cause atria1 fibrillation was higher for the right than for the left atrium. These electrophysiological differences were associated with a higher incidence of fibrillation episodes when the stimulus was applied on the left atrium rather than the right one. Anatomical differences, possibly including different contributions of vagal and sympathetic impulses to the two atria or hemodynamic differences (contraction-excitation feedback?) might have played some role, but any hypothesis on this matter would be conjectural. The episodes of sustained atria1 fibrillation were significantly more frequent during hypo than hyperglycemia and when elicited in the left atrium rather than in the right one. The relationship of sustained atria1 fibrillation to the initial condition (atrium in which it started) might suggest that this arrhythmia does not involve all the atria1 myocardium at once. For the fibrillation to become sustained, a critical fibrillatory mass may have to be achieved within a critical time affected perhaps by the location of the stimulus, the level of serum glucose and, possibly, other factors, Hemodynamic changes also induced by the onset of atria1 fibrillation might favor its maintenance via some sort of contraction-excitation feedback.“-19 Such changes might be different initially depending on the site where the arrhythmia starts. Two points in this study might be of clinical interest. Paroxysmal atria1 fibrillation is prone to begin at rest, during sleep,2o when hypoglycemia is likely, adding a sympathetic factor to the vagal one prevailing at night. It is well known2’ that both the sympathetic and vagal stimulation shorten the atria1 refractoriness, thus - increasing the tendency for atria1 fibrillation to occur. The clinical occurrence of atria1 fibrillation during hypoglycemia has already been shown.5 The second point concerns the importance of a critical time for the establishment of sustained

atria1 fibrillation. If similar observations apply in the case of human atria, new strategies should probably aim at a swift intervention, perhaps electrical, before the critical time has elapsed and the arrhythmia is fully established.

Conclusion The incidence of atria1 fibrillation and its persistence in the dog is higher when a stimulus is applied on the left rather than the right atrium and when under a hypo rather than hyperglycemic state. The increased incidence of atria1 fibrillation is associated with a shorter refractory period of the left rather than right atrium and with hypo rather than hyperglycemia. The hypothesis that a critical time must elapse before atria1 fibrillation is fully established may explain the effect of stimulus characteristics on the persistence of atria1 fibrillation.

References 1. Parish AE, Sugar SJN, Fazekas JF: A relationship between electrocardiographic changes and hypokalemia in insulin-induced hypoglycemia. Am Heart J 43:815, 1952 2. Petersen KG, Schluter KJ, Kerp L: Regulation of serum potassium during insulin-induced hypoglycemia. Diabetes 31:615, 1982 3. Brown MJ, Brown DC, Murphy MB: Hypokalemia from beta-2-receptor stimulation by circulating epinephrine. N Engl J Med 309:1414, 1983 4. Heine RJ, Van der Heyden EAP, Van de Veen EA: Response to human and porcine insulin in healthy subjects. Lancet 2:946, 1989 5. Rokas S, Mavrikakis M, Iliopoulou A, Moulopoulos S: Proarrhythmic effects of reactive hypoglycemia. PACE 15:373, 1992 6. Leak D, Starr P: The mechanism of arrhythmias during insulin-induced hypoglycemia. Am Heart J 63:688, 1962 7. Read RC, Doherty JE: Cardiovascular effects of induced insulin hypoglycemia in man during the Hollander test. Am J Surg 119:155, 1970 8. Shimada R, Nakashima T, Nunoi K et al: Arrhythmia during insulin-induced hypoglycemia in a diabetic patient. Arch Intern Med 144: 1068, 1985 9. Gale EAM: Hypoglycemia and human insulin. Lancet 2: 1264, 1989 10. Andrus EC, Carter EP, Wheeler HA: The refractory period of the normally beating dog’s auricle, with a note on the occurrence of auricular fibrillation following a single stimulus. J Exp Med 51:357, 1930

Atrial Fibrillation 11. Michelucci A, Padeletti L, Fradella GA: Atria1 refractoriness and spontaneous or induced atria1 fibrillation. Acta Cardiol 5:333, 1982 12. Carber AJ, Cryer PE, Santiago JV et al: The role of adrenergic metabolisms in the substrate and hormonal response to insulin induced hypoglycaemia in man. J Clin Invest 58:7, 1976 13. Santiago JV, Calarke WL, Shan SD et al: Epinephrine, norepinephrine, glucagon and growth hormone release in association with physiological decrements in the plasma glucose concentration normal and diabetic man. J Clin Endocrinol Metab 51:877, 1980 14. Rensma PL, Allessie MA, Lamers WJEP et al: Length of excitation wave and susceptibility to reentrant atria1 arrhythmias in normal conscious dogs. Circ Res 62: 395, 1988 15. Wydham CRC, Amat-Y-Leon, Wu D et al: Effects of cycle length on atria1 vulnerability. Circulation 55: 260, 1977 16. Allessie MA, Bonke FIM, Schopman FJG: Circus movement in rabbit atria1 muscle as a mechanism of

17.

18.

19.

20.

21.

and Glycemia

l

Vardas et al.

153

tachycardia. II. The role of nonuniform recovery of excitability in the occurrence of unidirectional block as studied with multiple microelectrodes. Circ Res 39: 168, 1976 Lab MJ: Contraction-excitation feedback in myocardium: physiological basis and clinical relevance. Circ Res 50:757, 1982 Sideris DA, Toumanidis ST, Kostis EB et al: Response of tertiary centers to pressure changes: is there a mechano-electrical association? Cardio Res 24: 13, 1990 Klein LS, Miles WM, Zipes DP: Effect of atrioventricular interval during pacing or reciprocating tachycardia on atria1 size, pressure, and refractory period: contraction-excitation feedback in human atrium. Circulation 82:60, 1990 Coumel P, Leclercq JF, Attuel P: Paroxysmal atria1 Iibrillation. p. 158. In Kulbertus HE, Olsson SB, Sclepper M (eds): Atria1 fibrillation. AB Hassle, Molndal, 1982 Katz AM: Physiology of the heart. Raven Press, New York, I977