Anxiolyticlike Effects of Atrial Natriuretic Peptide on Cholecystokinin Tetrapeptide–Induced Panic Attacks: Preliminary Findings

Background  Panic attacks induced by administration of cholecystokinin tetrapeptide (CCK-4) have been evaluated as a valuable tool to investigate the neurobiological mechanisms involved in panic anxiety. The rationale to study the effects of natriuretic peptides on the CCK-4 response is derived from observations that atrial natriuretic peptide (ANP) is released during panic attacks in humans and has anxiolyticlike actions in various animal models.

Methods  A double-blind, placebo-controlled design was conducted in 9 patients with panic disorder and 9 similar healthy control subjects. After pretreatment with an infusion of 150 µg of ANP or placebo in random order, each subject received 50 µg of CCK-4. Psychopathological parameters as well as physiological measures were sampled before and after CCK-4 administration.

Results  After pretreatment with ANP, the number of CCK-4–induced panic attacks decreased from 8 to 6 in patients and from 5 to 2 in controls. Acute Panic Inventory ratings were significantly reduced in patients after ANP vs placebo pretreatment. Infusion of ANP significantly curtailed the CCK-4–induced release of corticotropin in patients. Heart rate variability analysis indicated a sympathetic stimulation by CCK-4 that was inhibited by ANP in patients and controls.

Conclusions  The present study indicates that ANP exerts anxiolyticlike effects on CCK-4–stimulated anxiety attacks in patients with panic disorder. In addition, ANP produced an inhibition of the hypothalamopituitary-adrenocortical system and sympatholytic effects.

PANIC ATTACKS are characterized by paroxysmal episodes of intense anxiety or discomfort combined with various autonomic symptoms. Some experimental evidence supports the theory that panic attacks are generated by neuronal discharges in the brainstem,¹ which are caused by a hypersensitive suffocation monitor that produces false alarms.² Although several experimental provocation paradigms for panic attacks have been developed,³ it has recently been hypothesized that panic attacks are associated with an excessive release of corticotropin-releasing hormone (CRH) within the central nervous system (CNS). In rats, this hormone activates noradrenergic neurons in the locus coeruleus,⁴ and direct infusion of CRH into this structure causes significant behavioral arousal.⁵ Complementarily, anxiogenic effects are blocked by CRH receptor antagonists⁶; antisense targeting of CRH receptor messenger RNA⁷ elicits anxiolytic effects, and CRH-1 receptor–deficient mice display considerably lower anxiety levels than wild-type control mice.⁸ In patients with panic disorder, the hypothalamopituitary-adrenocortical (HPA) system shows changes that point to a temporary hypothalamic CRH hypersecretion.⁹

Surprisingly, although living through a panic attack represents an intense stressor, most of the studies on naturally occurring attacks or those stimulated by sodium lactate found no significant activation of the HPA system.^10-12 In contrast to other compounds,¹³ atrial natriuretic peptide (ANP) is the only peptide that inhibits HPA activity in humans at all regulatory levels of the system.¹⁴ In patients with panic attacks, infusion of sodium lactate produces a more rapid and pronounced increase in ANP¹¹ in comparison with healthy control subjects,¹⁵ which might in part explain the frequently observed quiescence of the peripheral HPA system.

Sympathetic responses during the panicogenic challenge with sodium lactate are also absent, which again can be explained by an inhibitory action of elevated plasma ANP levels.¹⁶ Most importantly, ANP and related peptides exert anxiolyticlike effects in rats.¹⁷ Hence, it can be surmised that ANP, which is found not only in the cardiac atria but also in various brain regions, might serve as a humoral feedback signal to confine the psychopathological and neuroendocrine sequelae of panic anxiety, possibly via inhibition of CRH-mediated brain circuits.¹⁸

To investigate both processes, the panicogenic stimulus should activate not only panic anxiety but also the HPA system, which is provided by cholecystokinin tetrapeptide¹⁹ (CCK-4). In parallel with its panicogenic actions, CCK-4 induces a significant release of corticotropin and cortisol.^20,21

To test the hypothesis that ANP has an anxiolyticlike profile, the effect of an infusion of ANP vs placebo during a CCK-4 stimulus was investigated in patients with panic disorder and healthy controls in a prospective, randomized, double-blind study. The aim of the study was to clarify in detail whether peripherally administered ANP modulates the psychological, endocrine, and autonomic responses to CCK-4.

Nine patients with a diagnosis of panic disorder with and without agoraphobia (6 women and 3 men; mean ± SD age, 33 ± 9 years; DSM-III-R diagnoses 300.01/300.21) and 9 similar controls (6 women and 3 men; mean ± SD age, 30 ± 5 years) were studied. Controls were recruited by means of advertisement and informed about the protocol. All patients and controls were given the Structured Clinical Interview for the DSM-III-R (German version)²² by a trained rater (M.K.). None of the patients had other Axis I diagnoses, and none of the controls had any psychiatric diagnosis. All subjects were physically healthy and had been without prescription and nonprescription drugs for at least 10 days. The subjects were given a thorough medical examination, including urinary drug screening. Furthermore, all subjects were not sleep deprived, they refrained from alcohol and nicotine (no more than 3 cigarettes per day), and they had no additional stressful life events during the 3 months before the study period (ascertained by interview). The protocol used was approved by the Ethics Committee for Human Experiments, Max Planck Institute of Psychiatry, Munich, Germany, and written informed consent was obtained from each participant before the investigation.

All subjects were studied in a supine position under video observation in a soundproof private room from 9 AM to 1 PM on 2 separate days. After subjects had had a standardized breakfast without caffeine ingestion at 8 AM, an intravenous cannula was inserted into a forearm vein of each arm at 9 AM. Each cannula was connected to a long catheter that ran through a soundproof lock to the adjacent laboratory. The cannulae were kept patent using isotonic sodium chloride solution (the blood-sampling cannula at an infusion rate of 50 mL/h and the infusion cannula at a rate of 1 mL/h). At 11 AM on both study days, each subject received as bolus injection CCK-4 (Clinalfa, Läufelfingen, Switzerland), 50 µg, dissolved in 10 mL of isotonic sodium chloride solution. From 10:40 to 11:10 AM in a prospective, randomized, double-blind design, all participants underwent infusion with isotonic sodium chloride solution, 30 mL (placebo condition), or isotonic sodium chloride solution, 30 mL, containing human α-ANP (Clinalfa), 150 µg. Four of 9 patients and 5 of 9 controls received placebo on the first study day. Blood samples (8 mL each) were drawn at 10, 10:15, 10:30, 10:40, 10:50, 11, 11:05, 11:10, and 11:15 AM, and thereafter every 30 minutes from 11:30 AM to 1 PM for the determination of hormonal concentrations. These samples were placed on ice, plasma was immediately separated, and specimens were stored at −80°C until analysis. Blood pressure was registered at blood sampling times with an automatic device. An electrocardiogram (ECG) was recorded continuously from 10:30 AM to 1 PM using commercially available equipment (Spectra Scan Model 263; DelMar Avionics, Irvine, Calif) with a standard cassette tape recorder.

The Acute Panic Inventory²³ (API), a DSM-III-R panic symptom checklist,²⁴ and a 100-mm visual analog scale (VAS) for anxiety were administered before CCK-4 injection at 10:55 and then at 11:10 AM to determine post hoc the peak level of provoked panic. The rater (M.K.) was unaware of the ANP or placebo administration. The criterion for the presence of a panic attack was an API total score exceeding 20 or an increment of at least 14 points above the preinfusion score, the presence of at least 4 DSM-III-R panic items, and an increase of at least 40 mm in the VAS. Furthermore, the patients were asked whether they had had a panic attack, and they should have reported similarities of at least 70% to their spontaneously occurring attacks. On the second study day, all subjects were asked which investigation was less frightening.

Plasma cortisol concentrations were determined using a commercially available radioimmunoassay (ICN Biomedicals, Carson, Calif). The detection limit was 0.8 nmol/L; intra-assay and interassay coefficients of variation for 55.2- and 110.4-nmol/L levels were less than 7%. Corticotropin concentrations were determined using an immunoradiometric assay (Nichols Institute, San Juan Capistrano, Calif). The detection limit for plasma corticotropin was 0.8 pmol/L, and the intra-assay and interassay coefficients of variation at 4.4 pmol/L were less than 8%. To measure ANP level, we adjusted a radioimmunoassay (Nichols Institute) to our requirements as described elsewhere.¹⁵ Minimum detectable amounts were 8 ng/L. The intra-assay coefficient of variation was 8.5%, and the interassay coefficient was less than 10%, calculated at plasma levels of 80 ng/L after extraction.

Stored analog ECG data were digitalized at a rate of 128 samples per second (R-R signal values, distance of 2 successive R spikes per second), and an arrhythmia analysis was performed in a prospective user-interactive mode (Stratascan; DelMar Avionics). All successive R-R intervals were sampled, excluding ectopic beats and artifacts. Single missing events were interpolated if no sinus reset had occurred; all other artifacts were rejected. The mean location (ML; average of values in each time interval) of the R-R interval values or average of heart rate (in beats per minute) was calculated for 1-minute segments. For spectral domain parameters, the artifact-free R-R signal was equidistantly scanned as a step function subtracting the mean signal values, and a fast Fourier transformation was performed in overlapping 1- or 3-minute segments. From the spectra, we calculated total power (PTOT) between 0.01 and 0.45 Hz (in milliseconds squared) as indicator of heart rate fluctuation and low- (LF; 0.01-0.05 Hz), mid- (MF; 0.05-0.15 Hz), and high-frequency power (HF; 0.15-0.45 Hz), each given as percentage of PTOT. The LF/HF ratios were also calculated. Details of analysis are given elsewhere.¹⁶

Differences in the measured quantities (psychopathometric scores, hormonal concentrations, and spectral parameters) between patients and controls, between the placebo and ANP conditions, and among baseline, infusion (10:40-11 and 11-11:10 AM), and postinfusion intervals (11:10-11:30 AM) were tested for significance by means of a multivariate analysis of variance (MANOVA) with repeated-measures design. To compensate for baseline differences, all considered quantities were normalized to their mean values at baseline before being used in MANOVAs. The influential factors in the MANOVAs were group (a between-subjects factor), treatment, and time (both within-subjects factors). For hormonal concentrations, the ML (average of values in each time interval) and area under the curve (AUC) values were used in this analysis. When significant main and interaction effects were found in the MANOVA, univariate F tests followed to identify variables that contributed significantly to these effects. Tests with contrasts were also performed to locate pairs of time intervals with significant differences in those variables, on which the factor time showed significant main or interaction effects. For testing the frequency distribution of panic attacks between both groups, Fisher exact tests were performed. To differentiate the incidence of frightening in the postprovocation interview between placebo and ANP conditions, McNemar tests were also performed. Furthermore, the stability of the hormonal concentration curves, ie, whether concentration curve values of one sample were continuously above those of another sample, was tested for significance with the matched paired Wilcoxon test by considering at each time the part of each sample with hormonal concentrations above the common medians. As a nominal level of significance, .05 was accepted. To keep the type I error at .05 or less, all post hoc tests (univariate F tests and tests with contrasts) were performed at a reduced level of significance (ie, adjusted α according to the Bonferroni procedure). (After finding a global effect of a factor, all subsequently performed post hoc tests concerning simple effects of this factor were performed at a corrected level of significance that was generally equal to .05 divided by the number of the corresponding tests. If the P values of the underlying statistics to these tests were less than the corrected level of significance, we declared a statistical significance. However, we denoted significance using P<.05 because statistical statements have to be made at the nominal level of significance, ie, α = .05, and not at the corrected level, which can differ from case to case and from effect to effect.) All measurements are expressed as mean ± SEM.

According to the DSM-III-R symptom checklist and respective criteria in API and VAS scores, 8 of 9 patients and 5 of 9 controls experienced a panic attack during the placebo condition after CCK-4 administration (frequency distribution between patients and controls was nonsignificant [P = .06], Fisher exact test). During the ANP condition, the number of subjects with panic attacks was reduced from 8 to 6 patients and 5 to 2 controls. A free postprovocation interview on the second study day showed that 8 of 9 patients and 7 of 9 controls identified the ANP condition as less frightening (McNemar test, P = .008 and P = .02 for patients and controls, respectively).

Analysis of variance with the normed scores of API and VAS revealed a significant group (F_15,2 = 8.67; P = .003) and time effect (F_15,2 = 54.40; P<.001) and a marginally significant treatment × time effect. Variables API and VAS scores contributed significantly to the aforementioned effects (univariate F tests, P<.05). During placebo condition, API scores increased in patients after CCK-4 administration from 9.7 ± 2.5 to 30.1 ± 3.7, and in controls from 2.2 ± 0.7 to 15.4 ± 2.7. Similarly, the VAS scores increased from 37 ± 6 to 89 ± 4 in patients and from 9 ± 3 to 58 ± 8 in controls. Between patients and controls, the severity of symptoms after CCK-4 administration was significantly different (tests with contrasts, API, F_1,16 = 8.77 [P = .009]; VAS, F_1,16 = 18.27 [P = .001]).

Concomitant with ANP infusion, the CCK-4–induced increase of API scores in patients from 7.2 ± 2.8 to 24.2 ± 3.4 was significantly curtailed compared with placebo condition (tests with contrasts, P = .02). In controls, no differences of CCK-4–mediated increases of API scores between conditions were seen (14.2 ± 2.7 after ANP vs 15.4 ± 2.7 after placebo). After ANP administration, the VAS scores were slightly curtailed compared with those for placebo (patients, 73 ± 7 vs 89 ± 4; controls, 53 ± 8 vs 58 ± 8).

The AUC values of the normed corticotropin and cortisol concentrations revealed a significant time effect (F_2,16 = 21.84 [P<.001]), a marginal group effect (F_2,16 = 3.09 [P = .07]), and a significant group × treatment × time interaction effect (F_2,16 = 5.52 [P = .02]) caused by variables corticotropin and cortisol (univariate F tests, P<.05). Subsequent tests with contrasts showed that in the interval 10:40 to 11 AM, no significant differences emerged between the ANP and placebo conditions for the AUC and ML values of corticotropin and cortisol secretion.

During the placebo condition, patients and controls showed a significant increase in corticotropin and cortisol levels after injection of CCK-4 (tests with contrasts, P<.05). The AUC and ML values were elevated significantly, and the increase of corticotropin was blunted in patients in comparison with controls (ML corticotropin at 11 to 11:20 vs 10:40 to 11 AM in patients, 7.0 ± 1.3 vs 4.5 ± 0.7 pmol/L; in controls, 10.0 ± 2.4 vs 4.8 ± 0.8 pmol/L [tests with contrasts, P<.05]).

During the ANP condition, the increase of corticotropin and cortisol levels was curtailed mainly in controls (ML corticotropin at 11-11:20 AM, ANP vs placebo, 8.1 ± 1.6 vs 10.0 ± 2.4 pmol/L; ML cortisol at 11-11:20 AM, ANP vs placebo, 350.4 ± 41.4 vs 433.2 ± 63.5 nmol/L). The secretion of corticotropin in patients was significantly lower than in controls at 11 to 11:20 AM and 11:20 AM to 1 PM (Figure 1; tests with contrasts, P<.05).

After injection of CCK-4, plasma concentrations of corticotropin (Figure 1) and cortisol remained lowered in patients until the end of the investigation. Addressing the stability hypothesis, we found that levels of both hormones showed on average continuously lower concentration values after infusion of ANP in patients and controls (Wilcoxon matched paired test, P<.05).

Atrial natriuretic peptide plasma concentrations were significantly lower in patients than controls at 10:40 AM (P = .04). After CCK-4 and concomitant placebo infusion, ANP plasma concentrations in patients rose by 30% (from 37 ± 5 ng/L at 10:40 AM to 48 ± 4 ng/L at 11:10 AM), and in controls by 13% (from 67 ± 16 to 76 ± 15 ng/L). In contrast to controls, the ANP plasma concentrations within the patient sample showed a significant time effect (F_2,7 = 5.87 [P = .03]), which could be attributed mostly to differences between the 10:40 and 11:10 AM samples (test with contrasts, P = .008).

Infusion of ANP resulted in similar increases in ANP plasma concentrations in patients and controls with a 4- to 5-fold increase compared with baseline.

Analysis of the mean heart rates normalized to the baseline mean values revealed a significant time (F_3,10 = 24.81 [P<.001]) and interaction effect (treatment × time, F_3,10 = 10.87 [P = .002]). After CCK-4 injection, a short-lasting increase in heart rate occurred in patients and controls, which was slightly but significantly enhanced in patients during coadministration of ANP (tests with contrasts, P<.05).

In patients, the spectral PTOT showed a fast decrease after CCK-4 and concomitant ANP infusions (significant differences in the total power scores between 10:40-11 and 11-11:10 AM) and a delayed decrease after CCK-4 administration and placebo infusion (significant differences in the total power scores between 11-11:10 and 11:10-11:30 AM; tests with contrasts, P<.05). Controls did not show any significant change of PTOT.

For LF, significant time effects emerged (F_3,10 = 7.21 [P = .007]) with a significant decrease in the interval 10:40 to 11 AM compared with baseline and thereafter an increase after CCK-4 administration. The MF and HF values showed marginally significant time effects (F_3,10 = 3.15 [P = .07] and F_3,10 = 3.13 [P = .07], respectively) and HF showed a significant interaction effect (treatment × time, F_3,10 = 6.14 [P = .01]).

Comparison of the LF/HF ratios (Figure 2) before and after CCK-4 administration demonstrated significant changes over time only during placebo infusion in both patients and controls (F_3,10 = 3.60 [P = .05]). In the intervals before (10:40-11 AM) and after CCK-4 injection (11-11:10 AM), a shift of HF to LF occurred in patients during placebo infusion, indicating an enhancement of sympathetic activity before CCK-4 administration. This shift was observed in controls after CCK-4 administration (ie, 11-11:10 AM).

Concomitant infusion of ANP largely reduced the CCK-4 effects on the LF/HF ratio, indicating an inhibitory effect of ANP on sympathetic activity. When testing the stability of the LF/HF ratios over time, we found significant differences between the 2 treatments (Wilcoxon matched paired test, P = .04), but not between patients and controls.

Infusion of ANP did not induce significant changes in diastolic or systolic blood pressure in patients or controls. In addition, the injection of CCK-4 was without any detectable significant effect.

The first major result of our preliminary study is that administration of ANP reduces the CCK-4–elicited panic reaction in patients with panic disorder and to a lesser extent in healthy controls. Panicogenic effects of CCK-4 have been confirmed in recent years to be consistent, reproducible, and related to the dose administered with an enhanced sensitivity of panic patients compared with healthy controls.^19,25 Accordingly, the dose of 50 µg of CCK-4 chosen for our investigation is sufficient to elicit panic attacks in patients and controls. Cholecystokinin tetrapeptide acts via cholecystokinin B-type (CCK-B) receptors in the CNS.²⁶ In recent studies, the short-term administration of benzodiazepines, β-blockers, and CCK-B receptor antagonists²⁷ and long-term treatment with imipramine hydrochloride and serotonin reuptake inhibitors have been found to be effective²¹ for inhibiting CCK-4–mediated panic attacks.

Atrial natriuretic peptide and its corresponding receptors are found within the CNS in areas known to regulate emotional states, such as the amygdala.²⁸ Indirect support for the anxiolyticlike effects of peripherally circulating ANP derives from studies during human pregnancy. Here, plasma ANP concentrations increase 3-fold at term²⁹ with a concomitant reduction in panic disorder symptoms and a postpartum reexacerbation.³⁰ Furthermore, present and recent studies in patients with panic disorder have consistently shown lower basal plasma ANP concentrations in panickers than in nonpanickers.^11,15

Direct evidence of anxiolyticlike effects of ANP comes from preclinical studies, despite their limited commensurability to human panic attacks. Intracerebroventricular injection of ANP and related fragments into animals has revealed anxiolyticlike effects in paradigms such as the elevated plus maze.³¹ Because central effects of peripherally released peptides on CNS functions have been reproducibly found in man and animals,^32,33 the observation of anxiolyticlike effects after peripheral application corroborates our findings.¹⁷ Direct effects of ANP on CCK-B receptors can be excluded,⁶ but the level of interaction of ANP with CCK-4–mediated effects remains unclear. Besides an antagonism of the CCK-4–mediated CRH release,^6,34 a modulation by neural afferents also has to be considered.³⁵

The second important finding of our preliminary study is that ANP inhibits the CCK-4–induced rise of corticotropin and cortisol levels. In contrast to most findings in spontaneous^36,37 and lactate-induced panic attacks,^10-12 the heterogenous group of panicogenic agents consisting of fenfluramine hydrochloride, yohimbine, and CCK-4 consistently activates the HPA axis.³ This activation has to be related to an acute hypothalamic release of CRH, which can be inhibited by ANP.^{6,13,15,38,39} The finding of a blunted corticotropin release after CCK-4 administration in patients compared with controls possibly has to be attributed to a chronic hypersecretion of CRH as evidenced in depression, posttraumatic stress,^40,41 and panic disorder.⁹ In addition, the lowered increase of cortisol level in patients after CCK-4 treatment surmises an enhanced feedback regulation of the HPA system as shown in posttraumatic stress disorder.⁴¹

As observed during lactate infusions, CCK-4 administration also induces a pronounced increase of plasma ANP levels in patients. A cardiac or hypothalamic release mediated by CRH^6,18,42,43 could contribute to this rise in plasma ANP level.

Besides the small sample size, a limitation of this study is that the endocrine status of patients before study could have been further evaluated to differentiate endocrine responsivity.

Another important finding of our study is the vegetative alterations induced by CCK-4, as indicated by spectral analysis of heart rate. Patients with panic disorder show an elevated baseline autonomic activity manifested by an increased heart rate, which is in line with previous reports.^44,45 The elevated autonomic activity seems to reflect an increased prechallenge arousal, since studies not performed under laboratory conditions have revealed no abnormality in cardiovascular parameters.⁴⁶ Administration of CCK-4 led to a very short and limited increase of heart rate in patients and controls, as reported recently.^9,25 Despite the increased basal activity in patients, a strong autonomic response to CCK-4–induced panic does not occur, which has also been shown for lactate-induced panic attacks.^16,46

Concerning spectral parameters, the increase in heart rate is followed by a prolonged reduction of the PTOT, indicating a decrease in heart rate variability. The analysis of the domain parameters showed an increase of sympathetic activity in the time intervals before and after CCK-4 administration. These findings provide further evidence of an increased prechallenge sympathetic activity in patients but not in controls. The prechallenge and the CCK-4–mediated sympathetic activation are almost entirely inhibited by ANP.

Our findings of vegetative alterations during panic are in line with investigations using lactate^44,46 that reported a parasympathetic response in panickers in contrast to controls. In humans during infusion of ANP,⁴⁷ an inhibition of sympathetic activity together with lowered total and HF power was found. These inhibitory effects were also demonstrated in our present study and in lactate-induced panic attacks.¹⁶

Preclinical data also support this idea. Sympathetic denervation and chemical sympathectomy inhibited the ANP release in rats,⁴⁸ whereas sympathetic activation increased ANP release in isolated hearts.⁴⁹ Vice versa, ANP injections into the nucleus tractus solitarii reduced the sympathetic tone.^50,51

Further support for mutual interactions between the autonomic nervous system and ANP release is provided by the finding that fibers from the locus coeruleus to hypothalamic ANP neurons can activate its release. Moreover, considerable amounts of ANP have been found in the locus coeruleus.⁵² This structure appears to play a key role not only in the regulation of the sympathetic nervous system but also in CRH release, and links these regulatory circuits to ANP. If ANP release during a panic attack contributes to the eventual termination of the attack via endocrine and neural feedback loops between the heart, the hypothalamus, and the locus coeruleus, it may prove worthwhile to develop and study long-acting ANP analogues for their possible antipanic efficacy.

Accepted for publication December 21, 2000.

The authors gratefully acknowledge the skillful technical assistance of Kristina Knaudt and Gisela Gajewski.

Corresponding author and reprints: Klaus Wiedemann, Department of Psychiatry and Psychotherapy, University of Hamburg, Martinistrasse 52, 20246 Hamburg, Germany (e-mail: wiedeman@uke.uni-hamburg.de).