| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From the Institute of Medical Psychology (U.S., E.G., K.K., H.-J.S.), Diabetes Research Institute (Y.S.-N., A.H., F.A.G), and Medical Institute of Environmental Hygiene (H.-W.B.), Heinrich-Heine-University Düsseldorf, Düsseldorf, Germany.
Address reprint requests to: Ursula Stockhorst, PhD, Institute of Medical Psychology, Heinrich-Heine-University Düsseldorf, P.O. Box 101007, D-40001 Düsseldorf, Germany. Email: ursula.stockhorst{at}uni-duesseldorf.de
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
METHODS: Twenty healthy male students were assigned either to group 1, which received insulin (0.035 IU/kg IV), or to group 2, which received IV saline on 4 consecutive days (acquisition). On day 5 (test), both groups were injected with saline. The CS was an olfactory stimulus. Blood glucose, serum insulin, plasma glucagon, plasma catecholamines, serum cortisol, and symptoms were repeatedly measured during each session.
RESULTS: In the test phase, group 1 reacted with a significantly larger decrease of blood glucose after presentation of the CS than group 2. Within group 1, a larger conditioned blood glucose decrease was associated with features that enhance classical conditioning (ie, intensity of the unconditioned response and intensity of the CS). Furthermore, in group 1, there was an increase of baseline insulin from day 1 to day 5 and a tendency for insulin reduction after CS presentation. Groups also tended to differ in cortisol and neuroglycopenic symptoms after CS presentation.
CONCLUSIONS: Conditioned effects in blood glucose are in accordance with the predictions. As a result of the exploratory analyses, our data also provide hints about conditioned changes in insulin, counterregulatory hormones, and symptoms.
Key Words: insulin • classical conditioning • humans • blood glucose • counterregulatory hormones • symptoms
Abbreviations: CNS = central nervous system; COND = conditionable; CR = conditioned response; CS = conditioned stimulus; CV= coefficient of variation; EDTA = ethylene diamine tetraaceticacid; EGTA = ethylene glycol-bis(beta-aminoethylether)N,N-tetraacetic acid; HPLC = high-pressure liquidchromatography; IU = international unit; IV = intravenous; MEIA = microparticle enzyme immunoassay; NA = noradrenaline; RIA = radio immunoassay; SEM = standard error of mean; UR = unconditioned response; US = unconditioned stimulus.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Until now, conditioning of insulin effects has been studied mainly in animals, with a focus on blood glucose as the dependent variable. The first data came from Russian laboratories (7), but control groups were lacking. Controlled experiments of classical conditioning of blood glucose in rats were initiated by Woods et al. (8). In most of their studies (reviewed in Ref. 4), a relatively high dose of insulin (50 IU/kg) was used. Rats were either exposed to CS–insulin pairings (experimental group) or to CS–saline pairings (control group). When receiving the CS, typically an olfactory stimulus (9), and a saline injection in the test, subjects in the experimental group with the previous CS–insulin experience responded to the CS with a reduction of blood glucose, defined as conditioned hypoglycemia (eg, Ref. 10). Several experiments supported the interpretation as a conditioning phenomenon (11, 12) and also revealed its vagal mediation: Conditioned hypoglycemia was abolished by vagotomy, atropine (13, 14), and streptozotocin, a drug that destroys pancreatic B cells (15). Conditioned hyperglycemia instead of conditioned hypoglycemia, was obtained when a novel environment activating the pituitary–adrenal–sympathetic stress system was used as the CS paired with insulin (13, 16–19). Corresponding to the vagal mediation, Woods et al. (15, experiment 2) showed that conditioned blood glucose reduction is based on a (neurally mediated) conditioned release of insulin. However, so far only two animal conditioning studies have simultaneously analyzed blood glucose and insulin (20, 21); however, in these studies, tolbutamide (20) and food (21), not insulin, were used as the nominal US. Thus, more data on insulin secretion in a classical conditioning protocol using insulin need to be obtained.
Classical conditioning of insulin effects in humans has rarely been studied. The first hints of conditioning effects were obtained from schizophrenic patients treated with insulin shock therapy (7): Blood glucose decreased and hypoglycemic symptoms (eg, drowsiness and sweating) occurred under saline after previous insulin injections in the same setting. However, methodological shortcomings (missing control groups) are obvious. More than 30 years later, experiments with healthy humans started. Fehm-Wolfsdorf et al. (22) initially used a differential conditioning design. In the test session, when the CS+ and the CS- were both presented with saline, a conditioned blood glucose reduction to the CS+ could not be detected. But in a second experiment using a two-group design (23) with two types of CS (a stimulus compound versus the injection procedure per se) functioning as the independent variable, substantial deviations (>10 mg/dl) from the baseline blood glucose level were found in 11 of 25 subjects (44%). Bidirectional changes occurred, with the majority of subjects (N = 8) responding to the CS with a glucose decrease, two with a rise, and one with fluctuations of 20 mg/dl. Salivary cortisol was measured, but a CR was not found.
Up to now, insulin conditioning experiments in healthy humans have not included a control group receiving placebo on all days of the conditioning protocol. Consequently, we used a placebo-controlled, two-group design. In addition to examining classical conditioning of the blood glucose response, we also wanted to obtain pilot data on the mediation of conditioned blood glucose decrease by conditioned insulin secretion and on the conditionability of counterregulatory hormones and symptoms, thus extending the scope of variables. These additional questions constitute the pilot part of our study. As pointed out above, even in animal conditioning studies, insulin was seldom measured. Furthermore, because level of blood glucose is not a simple UR to insulin but an integral result of several ongoing endocrine processes, measuring counterregulatory hormones is important. As Ramsay and Woods (2) emphasize, the centrally detectable effect of the drug, not the drug itself, constitutes the US. In the case of exogenously injected insulin, the US is the insulin detected by the insulin receptors in the CNS (4, 10).The actions the organism generates via the CNS constitute the true UR. The brain detects the fall of glucose (2, 24), initiates the secretion of counterregulatory hormones (glucagon and catecholamines in a first line of defense and then growth hormone and cortisol) (6, 25), and directly innervates organs (eg, pancreas and liver), which directly increase blood glucose levels (2). Because of its CNS-initiated release and neural control (3), the secretion of counterregulatory hormones should be conditionable. We know of only one animal study in which corticosterone was measured in addition to blood glucose (19); results revealed a trend for a conditioned corticosterone increase. A placebo-controlled experiment in which blood glucose, insulin, and counterregulatory hormones are all analyzed has not been published until now.
We hypothesized that human subjects who previously received CS–insulin pairings (experimental group) would show a larger decrease of blood glucose after CS presentation in the placebo test session than subjects who received the placebo in all sessions (control group). For exploratory purposes, we also addressed the mediation of conditioned hypoglycemia (13–15) by insulin and the conditionability of counterregulatory hormones and symptoms. We expected that the experimental group would react with a higher increase of insulin secretion in the test session than the control group. Based on the CNS-mediation of counterregulation, we expected a conditioned increase in glucagon, catecholamines, and cortisol and an increase in hypoglycemic symptoms in the experimental group as compared with the control group.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Based on the localization of insulin receptors in the olfactory bulbs (eg, Ref. 26), an olfactory stimulus served as the CS. It consisted of a mixture of 2.1 ml of rosewood oil (Oleum Rosea Ligni Brasil, Caelo, Caesar and Loretz, Hilden, Germany) and 1.4 ml of peppermint oil (Oleum Mentae Piperitae, 10%, Caelo, Caesar and Loretz) mixed in 50 ml of water (see below for application). The smell had been pretested for sufficient novelty, intensity, and pleasantness in an independent sample of healthy subjects. The CS was applied by a special device, which guaranteed that the smell was discretely applied during a predefined time interval and that the air in the room was not affected by the smell. Air flow was produced by a compressed-air tank equipped with a two-way magnet valve. Flow rate was controlled by a rotameter and set at 37.6 ml/second. This air flow was aromatized using the rosewood/peppermint oil emulsion, which was filled in a bubbling flask equipped with an asparger. Silicon tubes, fixed by nose goggles, conducted the air flow to the subjects’ nostrils. One minute before injection, air flow was initialized using an electronic timer and automatically ended 6 minutes later.
Subjects
Subjects were 20 healthy male students aged 20 to 30 years who were randomly assigned to group 1 (CS–INS) or group 2 (CS–NaCl). They were recruited on campus by an announcement for healthy, nonsmoking male students of normal body weight. Mean age (±SEM) was 25.2 ± 0.7 years in group 1 and 24.3 ± 0.7 years in group 2. Body mass index was within the normal range in both groups (mean = 22.7 ± 0.4 kg/m2 in group 1 and 23.1 ± 0. 5 kg/m2 in group 2). In addition to smokers (27), athletes (28) were excluded. Medical exclusion criteria were diabetes (also in parents or siblings), endocrinological and neurological diseases, cardiovascular disorders, and medication with glucocorticoids. Eligibility was determined during an interview and a physical examination. Before participation, subjects were informed about the study, in particular about the schedule, amount of blood loss expected during each session, and possible risks.
All subjects (future insulin and future saline) took part in a one-session prestudy in which they were injected with insulin (0.035 IU/kg) but did not receive any olfactory CS. The rationale for the prescreening with insulin was to identify those subjects that (1) do not reach a blood glucose minimum of at least 50 mg/dl (ie, hypoglycemia) under the insulin dose and/or (2) refuse to participate in the five sessions of the conditioning study because of side effects (eg, extreme hypoglycemia or venipuncture problems) or personal reasons. Of the 33 subjects approached in the prestudy, nine did not enter the conditioning study because of the first reason and three did not enter because of the second reason. One additional subject dropped out during session 2 of the conditioning protocol after experiencing extreme side effects. The mean interval between the prescreening and conditioning sessions was 76.4 days (SEM = 8.9 days). Degree of hypoglycemia at the time of maximum hypoglycemia (24 minutes after injection) during prescreening tended to be correlated with degree of hypoglycemia (minute 24) on day 1 of the acquisition phase (day 1, minute 24; r = 0.51, p < .10) in group 1 (CS–INS). The correlation attenuated on days 2 to 4, possibly indicating a separate influence of the new conditioning experience.
Subjects were instructed to fast overnight (for at least 8 hours with no food or beverages except water or unsweetened herbal tea), not to drink any alcohol during the total 5 days starting the evening before the first session, and not to exercise or participate in sports on the morning of the experiment. The subjects received a total payment of 300,00 Deutsche Marks (equivalent to about $170 US) at the end of the last session.
Experimental Procedure
Sessions were scheduled at 7:45 AM on 5 consecutive days. The time line of each session is summarized in Figure 1.
|
1 mm, Braun) was inserted into the antecubital vein of the nondominant arm. When local vein irritation occurred, the dominant arm was used the next day. The experimental procedure started at 8:00 AM. Two subjects were run in the same session with a 3-minute lag. To increase standardization and to guarantee the time schedule, the sequence of events during each session was presented on a computer screen. The person who performed the experiment completed each task and indicated its completion by responding on the computer keyboard, thus initiating the announcement of the subsequent task. Each session lasted 128 minutes and was divided into a preinjection phase (44 minutes, indicated as minute -44 to minute 0 in Figure 1) and a postinjection phase (84 minutes after injection). The first 20 minutes of the preinjection phase constituted the resting period, during which time subjects were allowed to read magazines. Twenty-four minutes before injection, a blood sample for analysis of blood glucose, insulin, glucagon, catecholamines, and cortisol was taken. Blood samples for glucose analyses were drawn from venous blood every 6 minutes. A total of 17 blood glucose samplings were obtained, four before (minutes -24, -18, -12, and -6) and 13 after application of the CS (starting with minute 0 immediately preceding the injection until minute 72). Samples for insulin and counterregulatory hormones were subsequently drawn every 24 minutes after injection (ie, at minutes 24, 48, and 72). Another blood sample for insulin determination was also obtained 8 minutes after injection (minute 8). Symptoms were recorded at six time points (minutes -24, 12, 24, 36, 48, and 72) using symptom lists. Each symptom list was presented in a printed version and contained 19 symptoms, including typical symptoms of hypoglycemia that had to be identified and rated by the subject (see Dependent Variables for details). Furthermore, 3 minutes after the smell was terminated (minute 8), subjects were given a questionnaire on the features of the CS smell and had to rate its pleasantness, novelty, and intensity on a nine-point rating scale. Finally (minute 81), subjects completed a questionnaire (substance perception) in which they indicated which substance they assumed they had received (insulin or saline).
Dependent Variables
Dependent variables were blood glucose, serum insulin, plasma glucagon, plasma adrenaline and noradrenaline, serum cortisol, and symptoms.
Blood glucose was determined from venous hemolysated blood samples (each 20 µl, filled into a 1.5-ml plastic container containing a hemolysis inhibitor) by an Eppendorf EPOS analyzer 5040 using an enzymatic hexokinase method. Each blood glucose analysis was done in duplicate. The result was expressed as the mean of the duplicate measurement. In case of a difference of 
6 mg/dl, a third measurement was performed, and the two less distant values were used to calculate the mean. The mean coefficient of retest reliability (averaged over the 17 blood samples for each group) ranged from 0.80 to 0.95 for days 1 to 5. Standard measurement error per day reached mean values between 1.4 and 1.8 mg/dl for the 5 days of the experiment. Samples for the later determination of insulin and counterregulatory hormones (glucagon, catecholamines, and cortisol) were stored on ice during each session. After the end of a session, the samples were immediately centrifuged (catecholamines: 10 minutes, 4°C; all other hormones: 10 minutes, room temperature) and frozen at -20°C until assays were performed. Serum insulin was assayed with a commercially available MEIA (Abbott, Wiesbaden, Germany), measuring biologically active immunoreactive human insulin. The CV of the method is 6.0% for a mean insulin level of 8.3 mU/liter, corresponding to the fasting insulin level, and 4.4% for mean insulin levels of 40.4 and 121.7 mU/liter. Glucagon was measured in Trasylol-EDTA plasma (1000 U/ml blood, Trasylol, Bayer, Leverkusen, Germany) using a RIA developed by Drost (29). This method has an interassay CV of 7.3% for mean glucagon values of 96 ± 7 pg/ml and 7.9% for mean glucagon values of 441 ± 35 pg/ml. Plasma catecholamines (adrenaline and noradrenaline) were analyzed by HPLC in EGTA-plasma using a Beckman System Gold Autosampler 507, a Pharmacia Beckman electrochemical detector, and commercial kits (Chromsystems, Munich, Germany). The method has an interassay CV for adrenaline of 5.6% for mean plasma adrenaline values of 64 and 5.8% for mean adrenaline values of 398. The interassay CV for noradrenaline is 6.1% for mean plasma noradrenaline values of 236 and 3.9% for mean noradrenaline values of 1315 pg/ml. Serum cortisol was measured using a commercially available RIA (Biermann, Bad Nauheim, Germany) with an interassay CV of 
6.4 for mean cortisol values ranging from 3.3 to 36.0 µg/dl.
The symptom list covered 19 symptoms: headache, palpitation, need to urinate, thirst, hunger, weakness, sweating, fatigue, dizziness, irritation, reduced concentration, anxiety, nervousness, trembling, blurred vision, nausea, salivation, breathing difficulty, and tingling around lips. Symptoms were identified and rated by the subject. Subjects were instructed to mark the actual intensity of each symptom by crossing a number on a five-point, numerically graded rating scale. The scale was also verbally anchored (0 = not at all, 1 = hardly, 2 = moderately, 3 = strong, and 4 = very strong). On the basis of results of hypoglycemia studies (eg, Refs. 30 and 31), symptoms were divided into neuroglycopenic and autonomic symptoms. The neuroglycopenic symptom score consisted of the symptoms weakness, fatigue, dizziness, blurred vision, and reduced concentration (see Ref. 31 for an analogous operationalization), which suggest a reduced glucose supply within the brain. The autonomic score consisted of sweating, tremor, palpitation, hunger, anxiety, and nervousness (cf. Ref. 31), indicating hypoglycemia-induced activation of the autonomic nervous system. The subjects’ responses on the symptom list were dichotomized into the values 0 (no) if the symptom intensity was qualified as "not at all" and 1 (yes) for intensity 1 to 4. Thus, the neuroglycopenic score, consisting of five symptoms, reached values between 0 and 5; the autonomic score (six symptoms) ranged from 0 to 6. In addition, all symptoms were added for a total symptom score (range, 0–19).
Data Analysis
Data are expressed as mean ± SEM. Because of a large variation between subjects in the baseline values of most of the endocrine parameters, differences from baseline values were calculated. Intergroup comparisons were made using the change scores. Inferences were made only for the cumulative blood glucose change on day 5 (test phase). Because of the small sample size, and to reduce the impact of outliers, comparisons were calculated with nonparametric tests for rank-scaled data. Accordingly, intergroup comparisons were calculated using the Mann-Whitney U test. Analyses were conducted with the SPSS statistical package. Because of the sample size (10 subjects per group), there were fewer than 30 different values in the combined sample; thus, reported p values are based on the calculation of exact tests. For intragroup comparisons, the Lam-Longnecker test for paired data (32) was used. Data analyses for insulin, counterregulatory hormones, and subjective symptoms are considered exploratory, not confirmatory, and were done for descriptive purpose only to address the pilot questions of the study (see Introduction). Accordingly, the latter p values were not corrected for multiple comparisons. Furthermore, we also report results reaching a p value of <.10 in the test session. The rationale is twofold. First, this is the first insulin conditioning study in which blood glucose, insulin, counterregulatory hormones, and symptoms were simultaneously measured. Thus, exploratory data must be gathered to identify parameters that are sensitive to conditioning. Second, even in the URs, intergroup differences did not attain the 5% but only the 10% level on some days during the acquisition phase. This is the case for cortisol and neuroglycopenic symptoms. It is reasonable to expect that the effects seen in the CR do not exceed those seen in the UR.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Blood Glucose
Figure 2 summarizes the blood glucose data during the acquisition phase (days 1–4) for group 1 (CS–INS) and group 2 (CS–NaCl). The blood glucose minimum in group 1 was obtained 24 minutes after injection of insulin and reached the level of hypoglycemia (50 mg/dl) on all 4 days (mean = 43.0 ± 3.1, 41.8 ± 3.3, 40.4 ± 2.6, and 43.4 ± 2.5 mg/dl on days 1–4, respectively). Group 2 reached mean values of about 75 mg/dl in the corresponding measurement point (minute 24) on all 4 days (mean = 75.5 ± 1.4, 76.3 ± 2.0, 75.0 ± 1.5, and 76.7 ± 1.2 mg/dl on days 1–4, respectively).
|
|
To describe the time course of blood glucose changes, U tests were performed for the difference from baseline for each measurement point. There was no hint of any group difference in the four pre-CS measurement points (minutes -24, -18, -12, and -6). But after CS presentation, group 1 reached lower values in 12 of the 13 post-CS measurement points, with p < .05 in three comparisons (minutes 18, 36, and 48) and p < .10 in four comparisons (minutes 0, 12, 42, and 66), indicating a systematic effect in the single measurement points (Figure 3, bottom).
Features of conditionability.
To explore whether a larger blood glucose decrease during the test is associated with variables that facilitate classical conditioning per se (ie, intensity of the CS and UR) (34) and thus validate the conditioning interpretation and to identify responder variables that enhance conditioning, we conducted the following post hoc analysis. Group 1 was divided into two subgroups on the basis of the maximum amount of blood glucose decrease after CS presentation. A median split resulted in five subjects (denoted as COND+) with a blood glucose decrease of >6 mg/dl on at least one time point after CS presentation and five subjects (COND-) with decreases of 6 mg/dl. The characteristics of COND+ and COND- subjects are summarized in Table 1. Because of the small sample size, descriptive data are given as medians.
|
Insulin.
Insulin injection resulted in the expected high insulin levels at the first postinjection measurement (minute 8, p = .000). Clear intergroup differences (p < .001) remained at minute 24 on all 4 days of the acquisition phase (Figure 4, left). Intergroup differences were no longer obvious at minutes 48 and 72. On days 3 and 4, comparisons revealed an even larger decrease for group 1 than for group 2 at minute 72 (day 3: U = 23.5, p = .022; day 4: U = 26.5, p = .067).
|
Counterregulatory hormones.
Results for the counterregulatory hormones glucagon, adrenaline, and noradrenaline are summarized in Table 2, and cortisol data are shown in Figure 4.
|
Adrenaline.
Intergroup differences in adrenaline during the acquisition phase peaked at minute 48 (day 1: U = 10, p = .003; day 2: U = 11.5, p = .001; day 3: U = 14.5, p = .003; day 4: U = 17, p = .011). At day 5, there were no effects in the difference-to-baseline scores.
Noradrenaline.
Noradrenaline effects were weaker and fluctuating. At day 1, an early (minute 24) increase in group 1 (U = 15.5, p = .012) was found. At day 3, the peak was delayed (minute 48), and groups only tended to differ (U = 32, p = .095). No difference was found at days 2 and 4. Accordingly, no CR was detected at day 5. Baseline levels at day 5 were lower than at day 4 but then rose in both groups, mainly from minutes 48 to 72 (p < .01, Lam-Longnecker-test).
Cortisol.
Our study was conducted in the morning (8:00–10:15 AM), when cortisol levels are decreasing (35). Thus, groups initially (minute 24) exhibited this circadian rhythm. But at minute 48, an opposite trend was observed: group 1 tended to have a lower decrease than group 2 (U = 31, p = .083) at day 1 that was more obvious at days 2 (U = 26.5, p = .038) and 3 (U = 21; p = .014), whereas p was only .062 at day 4 (U = 29). At day 5, groups again tended to differ at minute 48 (U = 32.5, p = .095). Baseline cortisol decreased from days 1 to 5 (group 1: Z = -6.354, p = .000; group 2: Z = -2.437, p = .007, Lam-Longnecker test for paired data).
Symptoms.
Figure 5 summarizes the change scores of the subjective symptoms, classified as neuroglycopenic symptoms, autonomic symptoms, and total number of symptoms, on days 1 to 5. Baseline values did not differ between groups at any day.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
It is obvious that the conditioned decrease of blood glucose in the experimental group compared with the control group is not large, also with regard to the measurement error of the blood glucose assay. But several points support the relevance and consistency of our results. In accordance with the majority of results obtained in animal experiments (see Ref. 4 for review), a blood glucose decrease, not an increase, was found in all subjects of our previous CS–insulin group. The fine-grain analysis of the post-CS measurement points had revealed a systematically lower level for group 1 compared with group 2 in 12 of the 13 measurement points. This clearly differs from the random behavior in the four pre-CS measurement points. Thus, while small in absolute terms, the decrease was present at almost every time point and thus differed from the pre-CS pattern. Thus, the results are more homogeneous than the response pattern obtained in a previous study with humans (23). Furthermore, the additional exploratory analyses (see below) are in accordance with the predictions.
To validate the conditioning interpretation, we conducted post hoc analyses, in which we again used blood glucose because it is the commonly used indicator of conditioning effects after the CS–insulin experience. On the basis of the amount of glucose reduction after CS presentation (ie, the amount of the CR), group 1 was divided into a subgroup with a higher (COND+) and a lower (COND-) conditioned blood glucose reduction. As expected from basic conditioning research (eg, Ref. 34), COND+ (ie, more conditionable) subjects showed features corresponding to parameters enhancing classical conditioning, such as a larger unconditioned blood glucose decrease (ie, a larger UR) at the end of the acquisition phase and they rated the CS as more intense. Furthermore, only COND+, but not COND-, subjects showed neuroglycopenic symptoms, whereas subgroups did not differ in the number of autonomic symptoms. Because neuroglycopenic symptoms are indicative of CNS effects of hypoglycemia (eg, Refs. 30 and 31), and because classical conditioning always requires the CNS, subgroup differences in neuroglycopenic (not in autonomic) symptoms are in accordance with the assumed mediation of classically conditioned responses. Concerning hormones, a lower baseline noradrenaline level was found in the COND+ subgroup. Because central noradrenaline also has a blood glucose increasing effect (eg, Refs. 3 and 37), the lower noradrenaline baseline corresponds with features that should facilitate conditioned blood glucose decrease. Although this is a post hoc analysis, it validates the conditioning interpretation of our data. Furthermore, it reveals endocrine and subjective responder variables that can be used as screening variables in insulin conditioning studies.
For insulin, group 1 compared with group 2 tended to decrease relative to baseline in minute 8 of the test session. This trend must be interpreted with caution, but it does not necessarily contradict the expected conditioned increase in insulin secretion (20). The first insulin sample was drawn only 8 minutes after injection. An increase might have occurred before that time point, as has been shown in animals (15). But our data also contain evidence of a conditioned insulin increase. In group 1, baseline insulin increased (p < .05) from day 1 (mean = 3.6 mU/liter) to day 5 (mean = 4.2 mU/liter). Although small in absolute level, this is a mean percentage increase of 27.8% in group 1 versus one of only 6.4% in group 2. This is the most obvious insulin effect in our study, which is the first to measure insulin after an acquisition phase with exogenous insulin. Such a baseline change might also indicate conditioning (2). We now suggest recording vagal activity continuously (eg, by measuring heart rate variability) (38–40) to examine whether the CS triggers a vagally mediated insulin secretion.
For glucagon, adrenaline, and noradrenaline, the descriptively conducted intergroup comparisons did not reveal conditioned effects in the difference-to-baseline values. However, there was a tendency for a conditioned cortisol response manifesting at the time point of the previous UR. This corresponds to findings of the only animal conditioning study with insulin in which corticosterone and blood glucose were both measured (19). Cortisol is also of interest for insulin conditioning because of its interaction with central insulin. Whereas insulin in the CNS leads to a reduction of food intake and body weight (4, 41, 42), glucocorticoids lead to increases in food intake (43, 44). Central insulin also interacts with the anorexic effects of peripheral cholecystokinin, now in an agonistic way (45). Thus, conditioning studies should also address the CNS effect of insulin, its role in food intake (46), and its interaction with neuropeptides.
Whereas group 1 and group 2 differed predominantly in autonomic symptoms during acquisition, neuroglycopenic symptoms were more sensitive to conditioning effects in the test session. Our results are in accordance with recent data (47) demonstrating that the expectation of receiving insulin led to a higher number of neuroglycopenic but not autonomic symptoms in comparison to subjects who were also injected with insulin but told they were receiving saline. Our results complete this picture for a classical conditioning situation. Furthermore, in our experiment, neuroglycopenic symptoms turned out to be more sensitive to adaptation. From days 1 to 5, there was a tendency for a decrease of baseline neuroglycopenic, but not autonomic, symptoms within group 1, who had experienced antecedent hypoglycemia. Adaptation to antecedent hypoglycemia with blunted symptom perception has also been described in patients with insulin-dependent diabetes mellitus (eg, Ref. 48), in patients with insulin-producing tumors (eg, Ref. 25), and in healthy subjects experiencing reoccurring hypoglycemia (eg, Ref. 49). However, the results were never related to classical conditioning. It would be interesting to analyze the contribution of classical conditioning to this change in symptom perception. Attenuation of the UR with repeated insulin injections also occurred with cortisol, in which, similarly to the neuroglycopenic symptoms, a trend for a CR could be demonstrated in the test session. One interesting question arising from this observation is whether an endocrine or subjective parameter that is sensitive to experience during acquisition is more easily conditionable.
In addition to the basic research questions that still must be solved, especially for human subjects (eg, classical conditioning of insulin effects and moderator variables or analysis of the interaction between central insulin and peptides or hormones), in larger samples, future studies should also analyze the impact of classical conditioning on the diabetic patient.
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Received for publication July 7, 1998.
Revision received February 3, 1999.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
2-receptor binding, norepinephrine, and circulating glucose. Brain Res 1986; 384: 404–8.[Medline]
This article has been cited by other articles:
 |
|
 |
|
  S. Chu Olfactory Conditioning of Positive Performance in Humans Chem Senses, January 1, 2008; 33(1): 65 - 71. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
) and 2 (
). Mean (M) ± SEM baseline (BA) values are given. Left, Neuroglycopenic score. Middle, Autonomic score. Right, Total number of symptoms score. Mean ± SEM values are given. Results of descriptively conducted Mann-Whitney U tests: [*] = p < .10; * = p < .05; ** = p < .01; *** = p < .001.