Cardiac toxicities of antibiotics.

Isolated heart muscle preparations are useful in the study of cardiac toxicities of drugs and environmental chemicals: such tissues allow assessment of chemical effects on heart muscle that is free from indirect in vivo influences that can mask or even accentuate cardiac responses measured in the intact animal. In the present study, left atria of guinea pigs were used to demonstrate a direct cardiac depressant effect of greater-than-therapeutic concentrations of several aminoglycoside antibiotics. The toxic effect of these antibiotics seems to be a calcium-dependent event, and may prove useful to characterize contractile responses of the heart. Other antibiotic agents can also depress cardiovascular function, as summarized in this report, but mechanisms of action have not been clearly defined.


Introduction
A variety of pharmacologic and environmental chemicals can influence cardiovascular function; however, such effects are not always easily detectable before serious damage to the heart or blood vessels has occurred. The heart is particularly susceptible to the actions of exogenous chemicals due, in no small part, to the functional dependence of this organ on dynamic and discrete ionic balances at the myocardial cell level (1). Not only can drugs directly affect myocardial integrity, but they may also modify cardiac performance by decreasing the ability of the heart to compensate when affected by disease. In recognition of such problems and the resulting need for experimental means of identifying cardiovascular effects of chemicals, "model" cardiac systems have been developed to characterize normal heart function and to predict the responsiveness of this organ to chemical challenge. Although limitations are routinely encountered when effects of drugs in isolated tissue models are extrapolated directly to human populations, heart muscle preparations have generally proven to be useful in the study of cardiac toxicities of chemicals.
Recently, we became interested in cardiac muscle and vascular smooth muscle models and have found certain in vitro preparations to provide predictive * Department of Pharmacology, Southwestern Medical School, The University of Texas Health Science Center at Dallas, Dallas, Texas 75235. information relative to in vivo cardiovascular manifestations of drug toxicities (2)(3)(4). In addition, a particular group of clinically used agents was identified to also have potential application for probing cardiovascular function. These drugs, the aminoglycoside antibiotics (i.e., neomycin-streptomycin group), were found to depress arterial muscle function in rather specific manners that could best be explained by an inhibitory effect on a portion of cellular calcium ion (Ca2+) bound to superficial membrane sites (2,5,6). Due to the contractile dependence of heart muscle on adequate cellular stores of Ca2+ (1,7), it also seemed that the cardiac depressant effects of these agents could involve Ca2+. Studies with atrial muscle of rats supported this contention (3) but provided little information relative to the influence of these agents on the contractile reactivity of heart muscle to inotropic interventions. The present report describes preliminary studies of the effects of selected aminoglycoside antibiotics on contractile function of isolated heart muscle of the guinea pig and also summarizes various aspects of the general subject of cardiac toxicities of antibiotic agents.

Materials and Methods
Male albino guinea pigs weighing between 150 and 250 g were decapitated with a guillotine; the hearts were rapidly excised and immediately placed in a Krebs-bicarbonate buffered solution (compo-October 1978 217 sition given below) aerated with 95% 02-5% Co2. The tension developed by atria was recorded in a similar way to that previously reported (8). The left atrial appendage was carefully dissected from the heart, mounted between two miniature clamps and suspended vertically in a 15 ml tissue bath. The upper clamp was connected by silk suture to a precalibrated force-displacement transducer (Grass Ft .03) for measuring isometric contractions on a multichannel recorder (Grass model 7B). The lower clamp was connected directly to a stationary rod in the bottom of the tissue bath. Muscle length was adjusted by raising or lowering the forcedisplacement transducer mounting unit with an integral micromanipulator. Initially, length-tension relationships were tested with each muscle and 1 g of diastolic tension was found consistently to yield maximum or near maximum contractile tension. Thereafter, 1 g of diastolic tension was placed on each muscle. The maximum rate of tension development (d Tldt), a more sensitive inotropic index than tension alone, was measured in some preparations by electrically differentiating contractile tension by use of a Grass 7P20 differentiator.
Electrical stimulation was accomplished with single square wave impulses of greater-thanthreshold voltage and 2 msec pulse duration, delivered at designated frequencies (1.0 Hz unless stated otherwise) from a Grass S44 stimulator.
Contractile tension and d TIdt were measured and expressed as grams of developed tension (or force) and grams per second, respectively, or values obtained in the presence of a drug were expressed as a percentage of predrug (control) values obtained in that muscle. All measurements are expressed as the mean + 1 standard error of the mean (X + SE) and the difference between two means was evaluated statistically by Student's t test.
The drugs used were 1-isoproterenol hydrochloride (Sigma Chemical Company, St. Louis, Mo.), digoxin (The Vitarine Co., Inc., New York) and the sulfate salts of gentamicin (Schering Corp., New Jersey), kanamycin (Sigma), amikacin (BB-K8; Bristol Laboratories, New York) and sisomicin (Schering). All concentrations refer to the base. Effects of drugs on heart muscle were exam-ined by additions of small aliquots of concentrated stock solutions directly to the bathing medium of the tissue bath. After a drug response was obtained for the designated time interval, the bathing solution was changed several times and replaced with fresh medium. If the muscle was to be reexamined for paired responses, time was allowed for contraction to return to control values and to stabilize before proceeding with an experiment. With all preparations, the muscle was allowed to stabilize for at least 45 min after contraction had been initiated before an experiment was started, and bathing solution was routinely replaced every 15-20 min.

Results
Initial studies evaluated the effects of gentamicin (2.0, 4.0mM) on contractile responses to Ca2+. Cumulative concentration-response curves to Ca2+ (2.5-14.0mM) were first determined in atria not treated with gentamicin and then (after appropriate rinsings and restabilization) redetermined after the negative inotropic effects of gentamicin reached a steady-state value (about 15 min). As shown in Fig  Environmental Health Perspectives fects of gentamicin, the antibiotic-treated heart muscle preparations responded to the positive inotropic effects of CA2+. In fact, when the larger concentrations of Ca2+ (> 8mM) were tested, the contractile responses of gentamicin-treated atria were similar to control Ca2+ responses (Fig. 1). Thus, the cardiac depressant activities of even large concentrations of gentamicin could be overcome by increasing Ca2+ availability.
If the Ca2+-gentamicin interaction represented a simple physiologic antagonism, then positive inotropic agents other than Ca2+ should also be expected to overcome the depressant action of the antibiotic. This possibility was examined by comparing the contractile effects of different interventions (i.e., isoproterenol, digoxin and increased frequency of stimulation) in control atria and in atria treated with large concentrations of gentamicin.
Responses to cumulative increases in the concentration of isoproterenol (1 x 10-10 to 3 x 10-7M) were determined in eight control atria and in eight other atria treated either with 2mM (n = 4) or 4mM (n = 4) gentamicin. As shown in Figure 2, isoproterenol produced a characteristic increase in contractile tension and dT/dt in both control and gentamicin-treated tissues. However, contractile responses to isoproterenol were reduced in gentamicin-treated atria, and even large amounts of isoproterenol (> 1 x 10-8M) could not increase contractility of gentamicin-treated atria to the magnitude seen in control atria (Fig. 2).
Contractile responses to digoxin (1 to 4 ,tM) in control atria (n = 12) and in atria exposed to 1 (n = 5), 2 (n=4), or 4mM (n=3) gentamicin are summarized in Figure 3. Although gentamicin-treated muscles  demonstrated inotropic responsiveness to digoxin, contractile force of these tissues did not reach the level observed in control atria (Fig. 3). Data in Figure 3 also demonstrate that the effects of gentamicin in control atria (i.e., not exposed to any drug) are concentration-dependent since 1, 2, and 4mM reduced contractile strength by about 35, 55, and 80%, respectively, prior to exposure to digoxin. Similar concentration-dependent effects of gentamicin were observed over a wide range of concentrations (0.1-4mM) of the antibiotic (data not shown). Frequency-tension relationships were determined in atria before treatment with gentamicin and again after the negative inotropic effect of gentamicin reached steady state. As shown in Figure 4 with typical myograms and as summarized in Figure 5, contractility of gentamicin-treated atria increased as stimulation frequency was increased from 0.1 to 2.0 Hz. However, contractile tension and dTldt responses of these atria did not reach the values obtained by control atria (Figs. 4 and 5) Figures 4 and 5, since responses of the atria after gentamicin was removed from the tissue bath (recovery data) were similar to control data obtained in the pre-gentamicin period.
Although present studies utilized gentamicin as a representative aminoglycoside agent, several test experiments were conducted with sisomicin, amikacin and kanamycin. As was seen with gentamicin, large concentrations (> 2mM) of these l~agents were required to appreciably reduce contractile force. Absolute values of contractile tension of atria exposed to a selected concentration of each 1'0 15 2.0 25 of these antibiotics are summarized in Table 1.

Discussion
Metabolic byproducts of various microorganisms have been utilized for different biomedical purposes, in addition to their most useful application as antibacterial drugs. In this respect, antibiotics could be considered as fortuitous and natural "contaminants" of the environment. The aminoglycoside group of antibiotics, for example, have greatly enhanced the treatment of a wide range of infections caused by gram-negative organisms. Although the bactericidal effects of these drugs differ as to specific spectrum of activity, these agents produce similar adverse side effects, including toxicities of the auditory-vestibular apparatus and the kidney (9), neuromuscular blockade (10) and, in some incidences, cardiovascular depression ( Table 2). In the present study, cardiotoxicity of selected aminoglycoside antibiotics was demonstrated and furthermore, evaluated to include potential interrelationships with other inotropic interventions.
Large concentrations of gentamicin, kanamycin, amikacin, and sisomicin reduced isometric contractile tension of electrically driven left atria of guinea pigs. This tissue preparation has the advantage of allowing assessment of drug effects on heart muscle that is free from indirect in vivo influences which can mask or even accentuate cardiac responses measured in the whole animal. Using gentamicin as a representative agent, we found that this drug not only produced a negative inotropic effect in isolated heart muscle, but also decreased contractile responses to several positive inotropic interventions, i.e., isoproterenol, Ca2+, digoxin, and increased frequency of stimulation. However, gentamicin interacted with Ca2+ responses differently than it affected responses to the other agents. Depression of cardiac function by gentamicin could be completely antagonized by increased concentrations of Ca2+, but only partially by the positive contractile effects of increased stimulation frequency or supramaximal concentrations of isoproterenol or digoxin. This difference suggests that the cardiac depressant ef-Environmental Health Perspectives   fect of aminoglycosides is a Ca2+-dependent phenomenon that can be antagonized by increased concentrations of Ca2+ in the interstitium, but only partially antagonized by procedures that mobilize available Ca2+ (e.g., isoproterenol, digoxin, and increased frequency of stimulation) (1, 7). The dependence of myocardial contraction on an increase in the intracellular concentration of free Ca2+ is now unequivocal, and an influx of Ca2+ into the myofiber directly activates actin-myosin elements and/or causes additional release of Ca2+ from cellular sequestration sites (1,7). The precise molecular mechanisms and cellular loci involved in myocardial Ca2+ fluxes have not been precisely identified and are under intensive investigation in many laboratories. In view of the effects of gentamicin on myocardial contractile function, this antibiotic may prove useful as a pharmacologic tool to probe Ca2+-dependent inotropic events in heart muscle. The antagonistic interaction of aminoglycosides and Ca2+ has been recognized in a variety of mammalian tissues (5,6,(12)(13)(14), and present findings provide a basis for more detailed investigation of the influence of these agents on Ca2+dependent processes in contracting myocardium.
Although the present study focused on the aminoglycosides, the potential cardiovascular toxicities of other antibiotics should not be disregarded. For example, large doses of lincomycin have been reported to disrupt impulse conductance through the excitable tissues of the myocardium, resulting in arrhythmias, cardiac standstill and, in some cases, ventricular fibrillation (15). Also it seems that the arrhythmogenic effects of digitalis can be affected by lincomycin (15). Hypotensive episodes have occurred during administration of chloramphenicol, and studies with perfused heart preparations revealed a myocardial depressant effect of this antibiotic (16). Studies with anesthetized dogs showed that tetracycline, vancomycin, erythromycin, and colymycin decreased cardiac output, systemic blood pressure and myocardial contractile force (17). Tetracyclines, erythromycin, oleandomycin, chloramphenicol, and colymycin have all been reported to decrease contractile force of isolated cardiac preparations (16)(17)(18). Table 2 summarizes several reports describing in vivo and in vitro cardiovascular effects of various antibiotics, including the aminoglycosides. Since many of these reports involve experimental studies with greaterthan-therapeutic doses of the antibiotics, these agents should not be indiscriminately categorized as Environmental Health Perspectives "cardiovascular depressants." Nevertheless, in view of the now rather lengthy list of antibiotics that have been reported to affect hemodynamics, the potential for cardiovascular toxicity with these agents should be considered when untoward circulatory changes are detected in patients undergoing antibacterial therapy.