Pharmacological preconditioning: comparison of desflurane, sevoflurane, isoflurane and halothane in rabbit myocardium

BACKGROUND

Recent investigations showed that isoflurane can induce pharmacological preconditioning. The present study aimed to compare the potency of four different halogenated anaesthetics to induce preconditioning.

METHODS

Anaesthetized open-chest rabbits underwent 30 min of coronary artery occlusion followed by 3 h of reperfusion. Before this, rabbits were randomized into one of five groups and underwent a treatment period consisting of either no intervention for 45 min (control; n = 10), or 30 min of 1 MAC halogenated anaesthetic inhalation followed by 15 min of washout. End-tidal concentrations of halogenated agents were 3.7% for sevoflurane (n = 11), 1.4% for halothane (n = 9), 2.0% for isoflurane (n = 11), and 8.9% for desflurane (n = 11). Area at risk and infarct size were assessed by blue dye injection and tetrazolium chloride staining.

RESULTS

Mean (SD) infarct size was 54 (18)% of the risk area in untreated controls and 40 (18)% in the sevoflurane group (P > 0.05, ns). In contrast, mean infarct size was significantly smaller in the halothane, isoflurane, and desflurane groups: 26 (18)%, 32 (18)% and 16 (17)%, respectively (P < 0.05 vs control).

CONCLUSIONS

Halothane, isoflurane and desflurane induced pharmacological preconditioning, whereas sevoflurane had no significant effect. In this preparation, desflurane was the most effective agent at preconditioning the myocardium against ischaemia.

Tissue Doppler imaging differentiates transmural from nontransmural acute myocardial infarction after reperfusion therapy

BACKGROUND

The evaluation of transmural extent of necrosis after acute myocardial infarction remains a major problem in clinical practice. We sought to determine whether color M-mode tissue Doppler imaging (TDI) could differentiate transmural from nontransmural myocardial infarction.

METHODS AND RESULTS

Twenty-one anesthetized open-chest dogs underwent 90 or 120 minutes of left anterior descending coronary artery occlusion followed by 180 minutes of reperfusion. The transmural extension of infarct was measured by triphenyltetrazolium chloride (TTC) staining. Segment shortening in the endocardium and epicardium of the anterior and posterior walls was assessed by sonomicrometry. Regional myocardial blood flow was measured by radioactive microspheres. TDI was obtained from an epicardial short-axis view. We calculated systolic and diastolic velocities within the endocardium and epicardium of myocardial walls and the subsequent myocardial velocity gradient (MVG). TTC staining could identify 2 groups according to the transmural extent of necrosis: 15 dogs had a nontransmural (NT) necrosis (42+/-3% of wall thickness), and 6 dogs developed a transmural (T) infarct (81+/-4% of wall thickness). In both groups, ischemia resulted in a significant and similar reduction in endocardial and epicardial velocities, with a resulting low systolic MVG in the anterior wall (0.10+/-0.07 in NT and 0.10+/-0.08 s(-1) in T). At 60 minutes of reperfusion, systolic MVG failed to change significantly in the transmural group (-0.20+/-0.09 s(-1)). In contrast, it increased significantly after reflow in the NT group compared with ischemic values (-0.99+/-0.20 versus 0.10+/-0.07 s(-1), P:<0.05).

CONCLUSIONS

TDI can differentiate transmural from nontransmural myocardial infarction early after reperfusion.

Brief preconditioning ischemia alters diacylglycerol content and composition in rabbit heart

In order to give further insight into the potential role of PKC in the beneficial effects of ischemic preconditioning, we have characterized the production of diacylglycerol, the endogenous activator of PKC, and its molecular species composition in ischemic control and preconditioned hearts. Preconditioning was induced by 1 cycle of 5 min of ischemia followed by 5 min of reperfusion. In control and preconditioned groups, hearts were harvested under deep anesthesia at baseline (preischemia) and at 2, 5 and 10 min into the sustained coronary artery occlusion, i.e., preceding myocyte death. Diacylglycerol content and fatty acid composition were analyzed by thin-layer chromatography (TLC) and high-performance liquid chromatography (HPLC), respectively. Myocardial diacylglycerol content was increased at 2 min into the sustained ischemia in the control group (481 +/- 34 vs 292 +/- 64 ng x mg(-1) at baseline; p < 0.05), but was comparable to the baseline value at 5 and 10 min. In the preconditioned group, diacylglycerol production remained unchanged throughout the 10-min test ischemia (317 +/- 17 at 2 min vs 312 +/- 38 ng x mg(-1) at baseline; p = NS). A detailed analysis of the molecular species composition at the time of 2 min revealed a reduced contribution of phosphatidylinositol to diacylglycerol production in preconditioned myocardium (global correlation coefficient 0.57 vs 0.66 in control myocardium) with a trend toward an enrichment of diacylglycerol composition with some species originating from phosphatidylcholine. Thus, our study revealed that brief preconditioning ischemia: (1) prevents the increase of diacylglycerol content in the early minutes of the sustained ischemia, and (2) emphasizes the contribution of phosphatidylcholine in diacylglycerol formation to the detriment of that of phosphatidylinositol.

Prevention of isoflurane-induced preconditioning by 5-hydroxydecanoate and gadolinium: possible involvement of mitochondrial adenosine triphosphate-sensitive potassium and stretch-activated channels

BACKGROUND

Both mitochondrial adenosine triphosphate-sensitive potassium (MKATP) channels (selectively blocked by 5-hydroxydecanoate) and stretch-activated channels (blocked by gadolinium) have been involved in the mechanism of ischemic preconditioning. Isoflurane can reproduce the protection afforded by ischemic preconditioning. We sought to determine whether isoflurane-induced preconditioning may involve MKATP and stretch-activated channels.

METHODS

Anesthetized open-chest rabbits underwent 30 min of coronary occlusion followed by 3 h of reperfusion. Before this, rabbits were randomized into one of six groups and underwent a treatment period consisting of either no intervention for 40 min (control group; n = 9) or 15 min of isoflurane inhalation (1.1% end tidal) followed by a 15-min washout period (isoflurane group; n = 9). The two groups received an intravenous bolus dose of either 5-hydroxydecanoate (5 mg/kg) or gadolinium (40 micromol/kg) before coronary occlusion and reperfusion (5-hydroxydecanoate, n = 9; gadolinium, n = 7). Two additional groups received 5-hydroxydecanoate or gadolinium before isoflurane exposure (isoflurane-5-hydroxydecanoate, n = 10; isoflurane-gadolinium, n = 8). Area at risk and infarct size were assessed by blue dye injection and tetrazolium chloride staining.

RESULTS

Area at risk was comparable among the six groups (29 +/- 7, 30 +/- 5, 27 +/- 6, 35 +/- 7, 31 +/- 7, and 27 +/- 4% of the left ventricle in the control, isoflurane, isoflurane-5-hydroxydecanoate, 5-hydroxydecanoate, isoflurane-gadolinium, and gadolinium groups, respectively). Infarct size averaged 60 +/- 20% (SD) in untreated controls versus 54 +/- 27 and 65 +/- 15% of the risk zone in 5-hydroxydecanoate- and gadolinium-treated controls (P = nonsignificant). In contrast, infarct size in the isoflurane group was significantly reduced to 26 +/- 11% of the risk zone (P < 0.05 vs.control). Both 5-hydroxydecanoate and gadolinium prevented this attenuation: infarct size averaged 68 +/- 23 and 56 +/- 21% of risk zone in the isoflurane-5-hydroxydecanoate and isoflurane-gadolinium groups, respectively (P = nonsignificant vs.control).

CONCLUSION

5-Hydroxydecanoate and gadolinium inhibited pharmacologic preconditioning by isoflurane. This result suggests that MKATP channels and mechanogated channels are probably involved in this protective mechanism.

Assessment of nonuniformity of transmural myocardial velocities by color-coded tissue Doppler imaging: characterization of normal, ischemic, and stunned myocardium

BACKGROUND

Transmural myocardial contractile performance is nonuniform across the different layers of the left ventricular wall. We evaluated the accuracy of color M-mode tissue Doppler imaging (TDI) to assess the transmural distribution of myocardial velocities and to quantify the severity of dysfunction induced by acute ischemia and reperfusion in the inner and outer myocardial layers.

METHODS AND RESULTS

Thirteen open-chest dogs underwent 15 minutes of left anterior descending coronary artery occlusion followed by 120 minutes of reperfusion. M-mode TDI was obtained from an epicardial short-axis view. Systolic velocities were calculated within endocardium and epicardium of the anterior and posterior walls. Regional myocardial blood flow was assessed by radioactive microspheres. Segment shortening was measured by sonomicrometry in endocardium and epicardium of both the anterior and posterior walls. At baseline, endocardial velocities were higher than epicardial velocities, resulting in an inner/outer myocardial velocity gradient. Ischemia caused a significant and comparable reduction in endocardial and epicardial systolic velocities in the anterior wall with the disappearance of the velocity gradient. Systolic velocities significantly correlated with segment shortening in both endocardium and epicardium during ischemia and reperfusion. In the first minutes after reflow, endocardial velocities showed a greater improvement than epicardial velocities, and the velocity gradient resumed although to a limited extent, indicative of stunning.

CONCLUSIONS

TDI is an accurate method to assess the nonuniformity of transmural velocities and may be a promising new tool for quantifying ischemia-induced regional myocardial dysfunction.

Opposite change with ischaemia in the antifibrillatory effects of class I and class IV antiarrhythmic drugs resulting from the alteration in ion transmembrane exchanges related to depolarization

It is known that class I antiarrhythmic drugs lose their antifibrillatory activity with severe ischaemia, whereas class IV antiarrhythmic drugs acquire such activity. Tachycardia, which is also a depolarizing factor, has recently been shown to give rise to an alteration of ion transmembrane exchanges which is particularly marked in the case of calcium. This leads one to wonder if the change in antifibrillatory activity of antiarrhythmic drugs caused by ischaemia depends on the same process. The change in antifibrillatory activity was studied in normal conditions ranging to those of severe ischaemia with a class I antiarrhythmic drug, flecainide (1.00 mg x kg(-1) plus 0.04 mg x kg(-1)x min(-1), a sodium channel blocker, and a class IV antiarrhythmic drug, verapamil (50 microg x kg(-1) plus 2 microg x kg(-1) x min(-1)), a calcium channel blocker. The experiments were performed in anaesthetized, open-chest pigs. The resulting blockade of each of these channels was assessed at the end of ischaemic periods of increasing duration (30, 60, 120, 180, 300, and 420 s) by determining the ventricular fibrillation threshold (VFT). VFT was determined by means of trains of diastolic stimuli of 100 ms duration delivered by a subepicardial electrode introduced into the myocardium (heart rate 180 beats per min). Ischaemia was induced by completely occluding the left anterior descending coronary artery. The monophasic action potential was recorded concurrently for the measurement of ventricular conduction time (VCT). The monophasic action potential duration (MAPD) varied with membrane polarization of the fibres. The blockade of sodium channels by flecainide, which normally raises VFT (7.0 +/- 0.4 to 13.8 +/- 0.8 mA, p < 0.001) and lengthens VCT (28 +/- 3 to 44 +/- 5 ms, p < 0.001), lost its effects in the course of ischaemia. This resulted in decreased counteraction of the ischaemia-induced fall of VFT and decreased aggravation of the ischaemia-induced lengthening of VCT. The blockade of calcium channels, which normally does not alter VFT (between 7.2 +/- 0.6 and 8.4 +/- 0.7 mA, n.s.) or VCT (between 30 +/- 2 and 34 +/- 3 ms, n.s.), slowed the ischaemia-induced fall of VFT. VFT required more time to reach 0 mA, thus delaying the onset of fibrillation. Membrane depolarization itself was opposed as the shortening of MAPD and the lengthening of VCT were also delayed. Consequently there is a progressive decrease in the role played by sodium channels during ischaemia in the rhythmic systolic depolarization of the ventricular fibres. This reduces or suppresses the ability of sodium channel blockers to act on excitability or conduction, and increases the role of calcium channel blockers in attenuating ischaemia-induced disorders.

Doppler tissue imaging quantitates regional wall motion during myocardial ischemia and reperfusion

BACKGROUND

Quantification of regional myocardial function is a major unresolved issue in cardiology. We evaluated the accuracy of pulsed Doppler tissue imaging (DTI), a new echocardiographic technique, to quantify regional myocardial dysfunction induced by acute ischemia and reperfusion.

METHODS AND RESULTS

In nine open-chest anesthetized pigs, various degrees of regional wall motion abnormalities were induced by graded reduction of left anterior descending coronary artery (LAD) blood flow. Pulsed Doppler tissue imaging was performed from an epicardial apical four-chamber view with the sample placed within the middle part of the septal wall. Peak septal velocities were calculated during systole, isovolumic relaxation, and early and late diastole. Regional myocardial blood flow and systolic and diastolic dysfunctions were assessed by radioactive microspheres and ultrasonic crystals, respectively. Ischemia resulted in a significant rapid reduction of systolic velocities and an early decrease in the ratio of early to late diastolic velocities. Both changes were detected by pulsed DTI within 5 seconds of coronary artery occlusion. The decrease in systolic velocity significantly correlated with both systolic shortening (r=.90, P<.0001) and regional myocardial blood flow (r=.96, P<.0001) during reduction of LAD blood flow.

CONCLUSIONS

These results suggest that DTI may be a promising new tool for the quantification of ischemia-induced regional myocardial dysfunction.

Stretch-induced protection shares a common mechanism with ischemic preconditioning in rabbit heart

We sought to determine whether stretch-induced preconditioning may be related to activation of adenosine receptors, ATP-sensitive K+ (K+ATP) channels, and/or protein kinase C (PKC) in the rabbit heart. Anesthetized rabbits underwent 30 min of coronary artery occlusion followed by 3 h of reperfusion. Ischemic preconditioning was induced by one episode of 5 min of ischemia followed by 5 min of reperfusion, and stretch preconditioning was induced by a transient volume overload. The abilities of gadolinium (Gd3+), a blocker of stretch-activated channels, glibenclamide (Glib), a blocker of K+ATP channels, 8-(p-sulfophenyl)-theophylline (8-SPT), a blocker of adenosine receptors, and polymyxin B (PMXB), an antagonist of PKC, to prevent the infarct size-limiting effect of stretch-induced preconditioning were evaluated. Because the infarct size-reducing effect of stretch occurred in the absence of ischemia and was prevented by previous administration of Gd3+, Glib, 8-SPT, and PMXB, we propose that activation of mechanosensitive ion channels protects the rabbit heart from subsequent sustained ischemic insult, likely through a mechanism that involves downstream activation of PKC, adenosine receptors, and/or K+ATP channels.

Ventricular fibrillation in preconditioned pig hearts: role of K+ATP channels

ATP-dependent potassium (K+ATP) channels play a role in the infarct size-limiting effect of preconditioning in pigs. We previously demonstrated that preconditioning shortens monophasic action potential duration (MAPD) and accelerates the time to ventricular fibrillation (VF) during a prolonged ischemia in pigs. We sought to determine whether the mechanism of the reduced time to VF in preconditioned pigs is a consequence of K+ATP, channel activation. Pigs underwent 40 min of coronary occlusion and 2 h of reperfusion. Before this, animals received either no intervention (control), 10 min of ischemia and 10 min of reperfusion (preconditioned), or an intravenous infusion of nicorandil, a K+ATP channel opener. Additional control, preconditioned, and nicorandil-treated pigs were pretreated by glibenclamide, an antagonist of K+ATP channels. Because 1) the K+ATP channel activator nicorandil did not produce shorter time to VF, 2) the K+ATP channel inhibitor glibenclamide did not block the acceleration of VF by preconditioning, and 3) there was no relationship between time to VF and infarct size or MAPD, the major conclusion is that reduced time to VF in preconditioned animals is not a consequence of K+ATP channel activation.

Cardiac beta-adrenoreceptor activation and ventricular fibrillation under normal and ischemic conditions

OBJECTIVES

To investigate the role of ventricular and atrial beta-adrenoceptor activation by isoprenaline in the genesis of rhythm disorders and risk of fibrillation in the healthy or ischaemic heart.

METHODS

The study was performed in anaesthetized, open-chest pigs. Electrical fibrillation threshold (EFT) of the ventricles was measured with trains of diastolic stimuli of 100 ms duration synchronized with respect to the R-waves and delivered to the myocardium by a subepicardial electrode introduced into the area which could be subjected to ischaemia. Monophasic action potential (MAP) and effective refractory period (ERP) were recorded in the same area. Ischaemia was obtained by complete occlusion of the left anterior descending coronary artery near its origin during increasing periods (30, 60, 90, 120, 150, 180, 240 s).

RESULTS

At a rate varying according to the action exerted by isoprenaline on the sinus rate, EFT decreased by about 30% in the healthy heart during the infusion of 0.5 micrograms/kg/min isoprenaline under the influence of the acceleration of cardiac beats. In the ischaemic heart, sinus tachycardia accelerated the fall in EFT and the reduction in MAP duration and resulted sooner in spontaneous ventricular fibrillation. During ventricular pacing at a constant rate of 200 beats/min, isoprenaline raised EFT by nearly 80% in the absence of ischaemia, but this rise was abolished by ischaemia, at least of no-flow type.

CONCLUSION

Tachycardia produced by activation of atrial adrenoceptors decreases EFT in the healthy heart and aggravates its fall in the ischaemic heart. Ventricular adrenoceptor activation counteracts the EFT fall related to tachycardia in the healthy heart, but not in the ischaemic heart. Therefore, the protection against ischaemic fibrillation due to beta-blockers would be essentially attributable to their action on the sinus nodes.

Change of a beneficial effect into an untoward effect by ischaemia: effect of quinidine-like drugs on vulnerability to ventricular fibrillation

The effects of three quinidine-like drugs, disopyramide, lidocaine and flecainide were investigated in anaesthetized, open-chest pigs on vulnerability to ventricular fibrillation under normal conditions and under myocardial ischaemia conditions. Vulnerability to fibrillation was evaluated by electrical ventricular fibrillation threshold (VFT), measured with 100 ms duration diastolic impulses the intensity of which was increased by steps of 1.0 or 0.5 mA. Impulses were delivered at the rate of 180 beats · min(-1). The ventricles were subjected to pacing at the same rate before the VFT determination, particularly throughout periods of ischaemia of increasing duration (30, 60, 90, 120, 150 s), separated by appropriate intervals for reproducibility of the results. Monophasic action potential (MAP) duration and conduction time were monitored in the ischaemic area under pacing. Ischaemia was obtained by complete occlusion of the left anterior descending coronary artery near its origin. The three drugs were i.v. administered in clinical dose range (1.00 mg · kg(-1) plus 0.04 mg · kg(-1) · min(-1)). In the absence of ischaemia, they increased almost equally VFT (from about 7 to 10 mA), despite 25% prolongation of conduction time. But, none of them was able to impede the increasingly marked fall of VFT caused by ischaemia: at 30 s, they had already lost any capacity for raising VFT and, beyond this time, they even aggravated its fall which led to spontaneous fibrillation when VFT approached 0 mA. The faster fall of VFT shortened time to onset of fibrillation (20 24 fibrillations for the three drugs at 150 s as against 12 24 in control period), the ischaemia-induced reduction of MAP duration (by 20%) being also hastened and slowing of conduction enhanced, given the addition of the depressant effects of ischaemia and drugs on conduction. Consequently, the antifibrillatory properties normally manifested by the studied drugs are first suppressed, then inverted by ischaemia.

[Value of calcium channel blockers in the prevention of ventricular fibrillation of ischemic etiology: experimental arguments]

The prevention of ventricular fibrillation raises a special problem when related to myocardial ischaemia, since class I antiarrhythmic drugs are then ineffective and may even behave as profibrillatory agents: the usual antifibrillatory properties of these drugs which are inhibitors of sodium channel, activated at high potentials, disappear with the disappearance of the role of sodium channel caused by ischaemic depolarization. Calcium channel then replacing sodium channel, calcium channel inhibitors should tend to prevent ischaemic ventricular fibrillation. Therefore, vulnerability to ventricular fibrillation was assessed in open-chest pigs by the threshold for fibrillation electrically induced with impulses of 100 ms duration at the rate of 180 beats/min. Ischaemia was produced by total occlusion of the left anterior descending coronary artery near its origin. Electrical fibrillation threshold was measured at the end of ischaemic period of increasing duration (30, 60, 120, 180, 240, 360 s) under control conditions and after i.v. administration of verapamil (50 micrograms/kg loading dose and 2 micrograms/kg/min infusion). Unaffected by verapamil when coronary circulation was normal, fibrillation threshold was raised by the drug when lowered by ischaemia, increasingly with the prolongation of ischaemia responsible for depolarization of the fibres, up to 500%. The rise of fibrillation threshold resulted in a delay in the triggering of fibrillation which occurs when the fibrillation threshold (6-8 mA) falls down to the pacing threshold (0.3-0.4 mA). These experiments tend to confirm the positive results recently obtained in man with verapamil in the prevention of postinfarction sudden death, provided that myocardial contractility is not too much adversely affected. But, in these experiments, left ventricular dP/dt max was not reduced by more than 15%, even just after the loading dose and returned to its control values within a few minutes.

Preconditioning reduces infarct size but accelerates time to ventricular fibrillation in ischemic pig heart

Preconditioning protects the rat heart from ventricular arrhythmias. However, the mechanism of this beneficial effect and its existence in large animal models remain unknown. We submitted 49 pigs to 40 min of left anterior descending coronary occlusion and 2 h of reperfusion and assessed the incidence of ventricular fibrillation (VF) and time to VF. Monophasic action potential duration (MAPD) and ventricular fibrillation threshold (VFT) were measured throughout the experiment. Preconditioning significantly reduced infarct size but failed to reduce the incidence of VF either during the 40-min ischemic insult or the following reperfusion. Moreover, preconditioning accelerated the onset of VF during the prolonged ischemia; time to VF averaged 8 +/- 2 min in the preconditioned group vs. 18 +/- 2 min in the control group (P < 0.05). This premature peak of VF in preconditioned hearts was associated with a significant decrease of VFT and shortening of MAPD. This suggests that preconditioning does not limit the incidence of VF in the pig model. Rather, preconditioning decreases the time to VF in this species, likely through lowering of the VFT and shortening of the action potential duration.

Bupivacaine hastens the ischemia-induced decrease of the electrical ventricular fibrillation threshold

Myocardial ischemia sensitizes the cardiotoxic effects of bupivacaine, especially the propensity to ventricular fibrillation. To investigate this sensitization and to elucidate its mechanism, the influence of bupivacaine alone, or associated with ischemia, was studied on electrical fibrillation threshold in anesthetized, open chest pigs. Determination of fibrillation threshold was performed with impulses of 100 ms duration at the rate of 180 bpm, in the absence of ischemia and at the end of increasing periods of ischemia (30, 60, 120, 180 s) obtained by complete occlusion of the left anterior descending coronary artery close to its origin. The effect of bupivacaine (1.00 mg/kg initial dose plus 0.04 mg.kg-1.min-1 over 25 min) was compared to the control in the same animals. This effect corresponded to 1.4-1.8 micrograms/mL plasma concentrations likely to be observed in humans after regional anesthesia. Bupivacaine significantly increased the fibrillation threshold before coronary occlusion from approximately 7.0 to 9.5 mA. In contrast, during ischemia the fibrillation threshold was shifted to the left and down, with a hastening of spontaneous fibrillation. Recording of monophasic action potentials in the ischemic area revealed that conduction time was prolonged by more than 100% under the combined influence of ischemia and bupivacaine, whereas the major enhancement of excitability due to ischemia was not attenuated by bupivacine. Therefore, bupivacaine should be used with caution in the condition of ischemia, especially if heart rate is rapid. In the present experiments, tachycardia is another factor in the enhancement of bupivacaine effects on conduction.

Possible role of drug interactions in bupivacaine-induced problems related to intraventricular conduction disorders

Clinically, bupivacaine has depressant effects on intraventricular conduction that may lead to serious atrioventricular blocks or reentrant arrhythmias at plasma levels below those required to produce these effects experimentally (2-3 micrograms/ml instead of 8-10 micrograms/ml). The difference could be due to drugs present in the blood at the time of regional anesthesia that similarly inhibit conduction. This hypothesis was examined in 30 anesthesized, closed-chest dogs by measuring conduction time in the ventricular contractile fibers as well as effective refractory period under pacing at a constant, relatively high (180 beats/minute) rate. Changes in sinus rate were limited, as well as changes in ventricular effective refractory period and blood pressure regardless of the drug tested. In contrast, cibenzoline, disopyramide, and propranolol increased conduction time and lengthened QRS duration. Clomipramine appeared to prolong conduction time and widen QRS only moderately in therapeutic doses, whereas verapamil did not manifest noticeable effects on conduction. Caution is therefore recommended in regional anesthesia with bupivacaine in subjects being treated with cardiovascular drugs, such as cibenzoline, disopyramide, and propranolol and their congeners, or even by tricyclic antidepressants.