- Email: [email protected]
Salivation and insulin secretion in response to food in non-obese men and women
Appetite, lournalfor Intake Research 1981,2,209-216
Salivation and Insulin Secretion in Response to Food in Non-obese Men and Women B. J. SAHAKIAN, M. E. J. LEAN M. R. C. Dunn Clinical Nutrition Centre, Cambridge
T. W. ROBBINS Department of Experimental Psychology, University of Cambridge
W. P. T. JAMES M. R. C. Dunn Clinical Nutrition Centre, Cambridge
Salivation and insulin secretion in response to the sight and smell of a palatable dinner of 14 non-obese subjects, who had not eaten for 17 h, were measured. There were significant increases in saliva and plasma insulin in response to the presentation offood, the increases in salivation correlating positively with indices of body weight (WIH2) and scores on the Herman and Polivy Scale of restrained versus unrestrained eating, but not with increases in insulin secretion. There were no changes in plasma glucose and glucagon over the time period examined. Interrelationships between these physiological changes and scores on other psychological tests, including the Stunkard and Messick Three Factor Scale and the Booth and Fuller Scale for emotional responsiveness, were also examined. Results are discussed in terms of possible extensions of these studies, and the significance of cephalic phase responses.
Cephalic reflexes are responses such as salivation, gastric acid and insulin secretion which appear to become conditioned to the anticipation of eating (see Powley, 1977). For example, Wooley and Wooley (1973) have shown that salivation occurred regularly when human subjects looked at palatable food. Furthermore, the amount of salivation increased in response to both dietary deprivation and palatability. The correlation coefficient of salivary output with hunger ratings showed r=0'75 and that for salivary output with appeal of food stimuli gave r=0·46. Recently, other investigators (Rodin, 1978; Sjostrom, Garellick, Krotkiewski & Luyckx, 1980) have demonstrated increased plasma insulin concentrations in response to food-related stimuli. Sjostrom et al. (1980) found a significant rise in insulin in response to food presentation in obese subjects, but not in non-obese subjects, while Rodin (1978) found that psychological traits, rather than body weight, were important correlates of the response.
We are very grateful to Mr Graham Jennings, Ms Sheila Bingham, Dr Claudia Sanchez-Castillo, and Ms Heather Davies for technical assistance. B. J. S. thanks Dr Diana Riad-Fahmy, Dr David Booth, Sir John Butterfield and Professor John Vallance-Owen for discussion. Requests for reprints should be sent to Dr B. J. Sahakian, M.R.C. Dunn Clinical Nutrition Centre. Addenbrooke's Hospital. Trumpington Street. Cambridge CB2 IQE. England. 0195-6663/81/030209 + 08 $02·00/0
(j
1981 Academic Press Inc. (London) Limited
210
B. 1. SAHAKIAN ET AL
On this basis, it is neither clear that insulin secretion in response to the sight and smell of food is a consistent phenomenon, nor what the behavioural or physiological significance of both salivation and insulin secretion might be. The present experiment was therefore planned to assess whether in food-deprived, non-obese subjects, salivation and insulin secretion in response to the sight and smell of palatable food would correlate with measures of hunger, food appeal, emotional responsiveness, eating behaviour, attitudes towards body weight and measured body weight. METHODS
Subjects
Table 1 gives relevant details of the 14 subjects studied. Seven of the subjects worked in the Departments of Anatomy or Experimental Psychology at the University of Cambridge and seven were undergraduates in the first year of the Experimental Psychology Course. All subjects selected had indicated by their ratings on preference scales that the food stimuli to be used, namely beef steak, fried chipped potatoes, onions, tomato and mushrooms, made a highly preferred lunch. The subjects, who were acquainted with each other, were requested to come for testing at luch time after not having eaten for 17 h. All subjects claimed to have complied with this requirement. The subjects were tested in a room which had been decorated to resemble the interior of a restaurant with posters on the wall, plants, flowers and an attractive table setting. The experimental atmosphere was kept as informal as possible without compromising experimental accuracy. The subjects were studied in three groups. TABLE 1 The sex, ({qe ({nd body weiqht index of'the subjects
Subject
1 2 3 4 5 6 7 8 9
10
11
12 13 14
Sex Male Male Male Male Male Male Male Male Male Male Female Female Female Female
Age (years)
W/H 2 (kg/m2)
20 21 20 26 20 46 19 26 34 21 20 34 20 22
18 21 22 23 24 24 25 25 25 26 17. 21 22 29
Note: Originally, there were 16 subjects; however, two subjects in a single group had to be eliminated, one for fainting shortly after inserting the venous cannula and the other for feeling faint on seeing his colleague's response. It was subsequently discovered that the first subject, despite volunteering for the study, had fainted several times in the past on venepuncture but had failed to proffer this information. The second subject displayed bradycardia, sweating and peripheral vasoconstriction and was eliminated from testing.
211
SALIVARY AND INSULIN RESPONSES TO FOOD
Procedure
On their arrival at the study centre, the subjects were asked to complete the scale devised by Herman and Polivy for distinguishing restrained from unrestrained attitudes and eating behaviour (Polivy, Herman & Warsh, 1978). The Booth and Fuller assessment for emotional responsiveness (Fuller, 1980) was also used. One week later, the subjects were contacted and asked to complete the Stunkard and Messick Three Factor Scale (Stunkard, 1981), where factors 1, 2 and 3 assess cognitive restraint, emotional lability and perceived hunger, respectively. The elapse of a week was used because seven of the ten questions on the Herman Polivy Scale are incorporated in the Stunkard and Messick Scale. The spacing of the questionnaires also limited any fatigue effects of multiple testing. Baseline measures of saliva output were taken. Six pre-weighed translucent cups were placed in front of each subject and the subjects were told to pick up the appropriately numbered cup and empty their mouths of saliva. The saliva was collected for signalled 2 min periods from 12-10, 8-6 and 4--2 min before the food stimuli were presented (pre-food stimuli saliva). Blood samples were withdrawn from an in-dwelling venous cannula inserted on arrival at the centre and before the questionnaires were administered. Sodium citrate solution was used to prevent blockage of the cannulae. Blood samples were taken at 10, 5 and 2 min before presentation of the food stimuli. The food was then immediately transferred from the adjacent kitchen into the experimental room and a plate of hot and steaming food was placed in front of each subject who was told that he/she would be allowed to eat the food shortly. The volunteers were then asked to rate their hunger on the Silverstone Analogue Scale (Silverstone, 1978), and to rate the appeal of the food on a similar analogue scale. Saliva samples were then collected at 0-2, 4--6 and 8-10 min after food presentation. Bloods were collected at 2, 4, 6 and 10 min following the arrival of the food and placed on ice immediately. During the collection period, subjects were informed that if they wished their food would be reheated for them before dining. After the last samples of saliva and blood had been collected, subjects were told that they could eat their lunches. All the food was consumed. Weight and height measurements were taken for each subject before the conclusion of the experiment. Immediately on conclusion of the experiment, saliva samples were weighed. Blood specimens were centrifuged at 4°e within 30 min, and plasma was stored at - 20 e until analysed for glucose, insulin and glucagon content. Glucose was measured by Beckman Analyser 2 and insulin and glucagon in duplicate by radioimmunoassay with CIS/INSIK 3 and Serono glucagon kits. 0
TABLE
2.
Saliva, insulin, glucose and glucagon before and cifter the presentation offood Saliva (g/2 min period)
Insulin (uU/ml)
Glucose (mmol/1)
Glucagon (pg/ml)
Pre food-stimuli
0'90±0'15
7'68±0'36
4·0± 1·0
194± 14
Post lood-stimuli
1'56±0'24
8·28 ± 0-42
4·0± 1·0
182± 17
t(13)=5'25 p<0'OO2
t(13) = 2-46 p<0'05
t(13) = 0·04 NS
t(13) = 1·57 NS
(Mean±SEM)
Paired t-test
212
B. J. SAHAKIAN ET AL
Statistical Analyses
Statistical analyses for changes in salivary output and blood measurements were made by paired Student's t-tests (Runyon & Haber, 1967). Correlations between the 14 subjects' scores on the different scales, the measures of increases in saliva and insulin, and the weight/height 2 indexes were examined by means of the Spearman Rank Correlation coefficient (rs) (Siegel, 1956). Where necessary, the corrected formula for use with tied observations was used, so as not to inflate the value of rs (Siegel, 1956, pp.206-21O). RESULTS
Table 2 shows that the group salivary output and plasma insulin concentration increased significantly after the food stimulus. This difference in average insulin values was minimized by including values up to 10 min after the food stimulus was presented (Figure 1). Comparing the reading immediately before, with the first reading after food presentation reveals a highly significant increase in both salivation (before: M ±SEM =0·95±0·17; after: M ± SEM = 1-42±0·23; p <0'002) and insulin secretion (Figure 1: t(13) = 5'31, p <0'002). There were no significant changes in either glucose or glucagon concentrations. As can be seen from Table 3,13 of the 14 subjects (p <0'001, binomial test) showed an increase in amount of saliva in response to the food stimulus and eleven (p < 0·029, binomial test) had increased insulin values. For glucose and glucagon changes, subjects showed no statistically significant direction of change.
(N=4)
100
1\
,
I,
95
~
~
I 90f
I
~(N=14)
I
85,
W~O'"
80[1
i
1No,,) \ 1NO"/,\
~, (N=14~
75
-10
\
(N=14)~
_S--~20,-+-'c2-~+--:6c--~+~IO
t
Food presentation Time (mini
FIGURE I. Plasma insulin concentrations (means) at different times before and after the presentation of food.
213
SAllY AR Y AND INSULIN RESPONSES TO FOOD
TABLE 3 The number of subjects showing an increase, decrease, or no change in saliva, insulin, glucose or glucagon in response to the presentation of food
Increase Post> Pre. No change Post= Pre. Decrease Post
Saliva
Insulin
Glucose
Glucagon
13
11
6
4
o
o
o
3
7
10
14
14
14
14
p<0'001
p<0'029
NS
NS
TABLE
4
Correlations between increases in salivation, body weight indices, and scores on the Herman and Polivy Scale Herman and Pol ivy Scale scores Increase in salivation W/H 2
rs= +0·60 (p
r,=+O'SS (p
* One-tailed test. There were statistically significant correlations between the degree of increase in salivation (mean pre-food stimuli values subtracted from mean post-food stimuli values), the body mass index and the Herman and Polivy Scale (Table 4). The degree of increase in insulin scores (mean pre-food stimuli values subtracted from mean postfood stimuli values) did not correlate significantly with the body mass index, the Herman and Polivy Scale or the salivary response. Neither the scores for food appeal nor scores on the Stunkard hunger rating scale correlated with the increase in salivation, in insulin or with any of the other measures. Seven of the ten questions on the Herman and Polivy Scale are also on the Stunkard and Messick Scale, thus allowing a test-retest correlation to be performed on those seven questions. The correlation between the first taking of the test and the second taking ofthe test, although highly significant (rs = + 0'87, p <0'01), became even greater with the removal of one subject's scores, those of subject 4 (rs= +0'94, p
214
B. 1. SAHAKIAN ET AL DISCUSSION
One of the main findings of this study was that non-obese men and women respond to the sight and smell of food by both salivating and increasing the insulin concentrations in peripheral blood. The salivary response has been documented in many studies by Wooley and Wooley (in press) who have used it as an index of involuntary appetite and observed increases in salivary output in response to food deprivation. The period of fasting in the current study was standardized to end at lunch-time after a 17 h fast so that the individual variability in preferences for eating at breakfast-time would not affect the responses and so that the standard food stimulus, previously tested for acceptability for all the subjects, would be considered appropriate for the time of day. This standardization may have helped to reveal the statistically significant relationships between the salivary response and both the subject's response rating in the Herman and Polivy questionnaire and his or her body mass index. The relationship of the salivary response to the body mass index would seem surprising given such a narrow weight range among the subjects, but this correlate was not distinguishable from the restraint scores since there was also a strong correlation between the body mass index and the degree of restraint. Five of the ten questions on the Herman and Pol ivy Scale are linked to awareness of weight changes. One explanation for these observations may then be that the relationship between salivation and body mass reflects predominantly a restraint effect rather than an innate link between the body's autonomic responsiveness, displayed by salivation and the body mass index. The alternative explanation that the salivary response in some way reflects the individual's ingestive behaviour and thereby the tendency to weight gain has been suggested (Wooley, Wooley & Woods, 1975), but is no more plausible an explanation for these findings than other suggestions. To distinguish between the effect of restraint and body weight would require the study of unrestrained individuals of widely different body weights and this type of study has not yet been performed. The findings in Table 2 of an insulin response to visual and olfactory cues from food is not new (Rodin, 1978), but does conflict with the careful studies of Sjostrom, Garellick, Krotkiewski and Luyckx (1980) who failed to find an insulin response in nonobese women. The change observed in the present study, as well as that from the study by Rodin (1978), is rather small. Careful triplicate analyses of plasma insulin are therefore necessary to display the effect and measurements need to undertaken 2-4 min after the stimulus has been applied if the effect is to be detected. Obese patients, with their insulin resistence and greater pancreatic secretion rates of insulin in response to a variety ofmetabolic or dietary stimuli, are more likely to have a detectable response to a neuronal effect mediated presumably via the vagus nerve. In the non-obese subject, therefore, great care is needed to ensure that the experimental procedures are undertaken with the minimum disturbance so that no adrenergic inhibition of insulin secretion is likely. Despite our efforts to minimize discomfort by the use of local anaesthesia for cannula insertion, two individuals in one group had to be eliminated from the analyses (Table 1). We were impressed by the need for careful organisation of blood taking and salivary collection in order to maintain an informal atmosphere, and consider this problem a potential error in studies of this nature. The other implication of the small plasma insulin response to food related stimuli is that this small increase in circulating insulin is unlikely in itself to mediate an increase in food intake in those with an enhanced response. Patients who are injected with
SALlY AR Y AND INSULIN RESPONSES TO FOOD
215
substantial amounts of insulin as part of insulin-hypo glycaemia tests rarely comment on hunger as a principal feature until the test is nearly over after 45-60 min and the blood glucose is beginning to rise. It seems unlikely therefore that the greater portal concentrations of insulin after neuronal activation could mediate an effect on hepatic receptors governing food intake. It seems more likely that insulin responses to food related stimuli, if eventually proven to correlate with eating behaviour, are not a primary link in the control of ingestive behaviour, but an index of behavioural responsiveness. The response did not, however, seem to correlate with analogue scores of hunger or of food acceptability.
CONCLUSION
To our knowledge, this is the first report of concurrent measurement of salivation and insulin secretion in response to food stimuli in the same subjects. The poor correlation within the group between the salivation and insulin responses suggests that these two responses, although affected by the autonomic system, need not be controlled by a common effector mechanism. We conclude that this system of testing for physiological correlates of appetite needs extending in a standardized fashion to subjects of different body weight with known responses to the current methods of assessing degree of eating restraint. There is also, however, a need to validate the use of these physiological tests to predict ingestive behaviour and then to explore whether the peripheral hormonal and gastrointestinal responses are among the mechanisms mediating an individual's eating pattern in response to visual and olfactory cues or merely an indirect manifestation of a central mechanisin. REFERENCES
Fuller. J. Human appetite and hodr si:=e control. the roles of indiridual differences and food dependencies in hU/11an appetite and hody size control processe,\. Unpublished doctoral dissertation, University of Birmingham, 1980. Polivy, J., Herman, C. P., & Warsh, S. Internal and external components of emotionality in restrained and unrestrained eaters. Journal of Abnormal Psychology, 1978,87,497-504. Powley, T. L. The ventromedial hypothalamic syndrome, satiety. and a cephalic phase hypothesis. Psychological Reriew, 1977,84,89-126. Rodin, J. Has the distinction between internal versus external control of feeding outlived its usefulness? In G. Bray (Ed .), Recent advances in obesity research (Vol. 2). Pp. 75-85. London: Newman, 1978. Runyon, R. P., & Haber, A. Fundamentals of behavioral statistics. Reading, Massachusetts: Addison-Wesley Publ. Co., 1967. Siegel, S. Non-parametric statistics for the behavioral sciences. London: McGraw-Hili, 1956. Silverstone, T. Drugs, appetite, and obesity in man. In G. Bray (Ed.), Recent advances in obesity research (Vol. 2). Pp.421-432. London: Newman, 1978. Sjostrom, L., Garellick, G., Krotkiewski, M., & Luyckx, A. Peripheral insulin in response to the . sight and smell of food. Metabolism, 1980,29,901-909. Stunkard, A. 1. "Restrained eating": what it is and a new scale to measure it. In L. A. Cioffi, W. P. T. James & T. Van Itallie (Eds.), Body weight regulatory systems: normal and disturbed mechanism. New York: Raven Press, 1981. Wooley, O. W., Wooley, S. c., & Woods, W.A. Effect of calories on appetite for palatable food in obese and non-obese humans. Journal of Comparative and Physiological Psychology, 1975, 89,619-625.
216
B. J. SAHAKIAN ET AL
Wooley. O. W .. & Wooley, S. C. Relationship of salivation in humans to deprivation. inhibition and the encephalization of hunger. Appetite: Journalfor Research Intake, 2, in press. Wooley, S. C., & Wooley, O. W. Salivation to the sight and thought of food: a new measure of appetite. Ps\'chosomatic Medicine, 1973,35, 136-142.
Received 1 June, 1981
Salivation and Insulin Secretion in Response to Food in Non-obese Men and Women B. J. SAHAKIAN, M. E. J. LEAN M. R. C. Dunn Clinical Nutrition Centre, Cambridge
T. W. ROBBINS Department of Experimental Psychology, University of Cambridge
W. P. T. JAMES M. R. C. Dunn Clinical Nutrition Centre, Cambridge
Salivation and insulin secretion in response to the sight and smell of a palatable dinner of 14 non-obese subjects, who had not eaten for 17 h, were measured. There were significant increases in saliva and plasma insulin in response to the presentation offood, the increases in salivation correlating positively with indices of body weight (WIH2) and scores on the Herman and Polivy Scale of restrained versus unrestrained eating, but not with increases in insulin secretion. There were no changes in plasma glucose and glucagon over the time period examined. Interrelationships between these physiological changes and scores on other psychological tests, including the Stunkard and Messick Three Factor Scale and the Booth and Fuller Scale for emotional responsiveness, were also examined. Results are discussed in terms of possible extensions of these studies, and the significance of cephalic phase responses.
Cephalic reflexes are responses such as salivation, gastric acid and insulin secretion which appear to become conditioned to the anticipation of eating (see Powley, 1977). For example, Wooley and Wooley (1973) have shown that salivation occurred regularly when human subjects looked at palatable food. Furthermore, the amount of salivation increased in response to both dietary deprivation and palatability. The correlation coefficient of salivary output with hunger ratings showed r=0'75 and that for salivary output with appeal of food stimuli gave r=0·46. Recently, other investigators (Rodin, 1978; Sjostrom, Garellick, Krotkiewski & Luyckx, 1980) have demonstrated increased plasma insulin concentrations in response to food-related stimuli. Sjostrom et al. (1980) found a significant rise in insulin in response to food presentation in obese subjects, but not in non-obese subjects, while Rodin (1978) found that psychological traits, rather than body weight, were important correlates of the response.
We are very grateful to Mr Graham Jennings, Ms Sheila Bingham, Dr Claudia Sanchez-Castillo, and Ms Heather Davies for technical assistance. B. J. S. thanks Dr Diana Riad-Fahmy, Dr David Booth, Sir John Butterfield and Professor John Vallance-Owen for discussion. Requests for reprints should be sent to Dr B. J. Sahakian, M.R.C. Dunn Clinical Nutrition Centre. Addenbrooke's Hospital. Trumpington Street. Cambridge CB2 IQE. England. 0195-6663/81/030209 + 08 $02·00/0
(j
1981 Academic Press Inc. (London) Limited
210
B. 1. SAHAKIAN ET AL
On this basis, it is neither clear that insulin secretion in response to the sight and smell of food is a consistent phenomenon, nor what the behavioural or physiological significance of both salivation and insulin secretion might be. The present experiment was therefore planned to assess whether in food-deprived, non-obese subjects, salivation and insulin secretion in response to the sight and smell of palatable food would correlate with measures of hunger, food appeal, emotional responsiveness, eating behaviour, attitudes towards body weight and measured body weight. METHODS
Subjects
Table 1 gives relevant details of the 14 subjects studied. Seven of the subjects worked in the Departments of Anatomy or Experimental Psychology at the University of Cambridge and seven were undergraduates in the first year of the Experimental Psychology Course. All subjects selected had indicated by their ratings on preference scales that the food stimuli to be used, namely beef steak, fried chipped potatoes, onions, tomato and mushrooms, made a highly preferred lunch. The subjects, who were acquainted with each other, were requested to come for testing at luch time after not having eaten for 17 h. All subjects claimed to have complied with this requirement. The subjects were tested in a room which had been decorated to resemble the interior of a restaurant with posters on the wall, plants, flowers and an attractive table setting. The experimental atmosphere was kept as informal as possible without compromising experimental accuracy. The subjects were studied in three groups. TABLE 1 The sex, ({qe ({nd body weiqht index of'the subjects
Subject
1 2 3 4 5 6 7 8 9
10
11
12 13 14
Sex Male Male Male Male Male Male Male Male Male Male Female Female Female Female
Age (years)
W/H 2 (kg/m2)
20 21 20 26 20 46 19 26 34 21 20 34 20 22
18 21 22 23 24 24 25 25 25 26 17. 21 22 29
Note: Originally, there were 16 subjects; however, two subjects in a single group had to be eliminated, one for fainting shortly after inserting the venous cannula and the other for feeling faint on seeing his colleague's response. It was subsequently discovered that the first subject, despite volunteering for the study, had fainted several times in the past on venepuncture but had failed to proffer this information. The second subject displayed bradycardia, sweating and peripheral vasoconstriction and was eliminated from testing.
211
SALIVARY AND INSULIN RESPONSES TO FOOD
Procedure
On their arrival at the study centre, the subjects were asked to complete the scale devised by Herman and Polivy for distinguishing restrained from unrestrained attitudes and eating behaviour (Polivy, Herman & Warsh, 1978). The Booth and Fuller assessment for emotional responsiveness (Fuller, 1980) was also used. One week later, the subjects were contacted and asked to complete the Stunkard and Messick Three Factor Scale (Stunkard, 1981), where factors 1, 2 and 3 assess cognitive restraint, emotional lability and perceived hunger, respectively. The elapse of a week was used because seven of the ten questions on the Herman Polivy Scale are incorporated in the Stunkard and Messick Scale. The spacing of the questionnaires also limited any fatigue effects of multiple testing. Baseline measures of saliva output were taken. Six pre-weighed translucent cups were placed in front of each subject and the subjects were told to pick up the appropriately numbered cup and empty their mouths of saliva. The saliva was collected for signalled 2 min periods from 12-10, 8-6 and 4--2 min before the food stimuli were presented (pre-food stimuli saliva). Blood samples were withdrawn from an in-dwelling venous cannula inserted on arrival at the centre and before the questionnaires were administered. Sodium citrate solution was used to prevent blockage of the cannulae. Blood samples were taken at 10, 5 and 2 min before presentation of the food stimuli. The food was then immediately transferred from the adjacent kitchen into the experimental room and a plate of hot and steaming food was placed in front of each subject who was told that he/she would be allowed to eat the food shortly. The volunteers were then asked to rate their hunger on the Silverstone Analogue Scale (Silverstone, 1978), and to rate the appeal of the food on a similar analogue scale. Saliva samples were then collected at 0-2, 4--6 and 8-10 min after food presentation. Bloods were collected at 2, 4, 6 and 10 min following the arrival of the food and placed on ice immediately. During the collection period, subjects were informed that if they wished their food would be reheated for them before dining. After the last samples of saliva and blood had been collected, subjects were told that they could eat their lunches. All the food was consumed. Weight and height measurements were taken for each subject before the conclusion of the experiment. Immediately on conclusion of the experiment, saliva samples were weighed. Blood specimens were centrifuged at 4°e within 30 min, and plasma was stored at - 20 e until analysed for glucose, insulin and glucagon content. Glucose was measured by Beckman Analyser 2 and insulin and glucagon in duplicate by radioimmunoassay with CIS/INSIK 3 and Serono glucagon kits. 0
TABLE
2.
Saliva, insulin, glucose and glucagon before and cifter the presentation offood Saliva (g/2 min period)
Insulin (uU/ml)
Glucose (mmol/1)
Glucagon (pg/ml)
Pre food-stimuli
0'90±0'15
7'68±0'36
4·0± 1·0
194± 14
Post lood-stimuli
1'56±0'24
8·28 ± 0-42
4·0± 1·0
182± 17
t(13)=5'25 p<0'OO2
t(13) = 2-46 p<0'05
t(13) = 0·04 NS
t(13) = 1·57 NS
(Mean±SEM)
Paired t-test
212
B. J. SAHAKIAN ET AL
Statistical Analyses
Statistical analyses for changes in salivary output and blood measurements were made by paired Student's t-tests (Runyon & Haber, 1967). Correlations between the 14 subjects' scores on the different scales, the measures of increases in saliva and insulin, and the weight/height 2 indexes were examined by means of the Spearman Rank Correlation coefficient (rs) (Siegel, 1956). Where necessary, the corrected formula for use with tied observations was used, so as not to inflate the value of rs (Siegel, 1956, pp.206-21O). RESULTS
Table 2 shows that the group salivary output and plasma insulin concentration increased significantly after the food stimulus. This difference in average insulin values was minimized by including values up to 10 min after the food stimulus was presented (Figure 1). Comparing the reading immediately before, with the first reading after food presentation reveals a highly significant increase in both salivation (before: M ±SEM =0·95±0·17; after: M ± SEM = 1-42±0·23; p <0'002) and insulin secretion (Figure 1: t(13) = 5'31, p <0'002). There were no significant changes in either glucose or glucagon concentrations. As can be seen from Table 3,13 of the 14 subjects (p <0'001, binomial test) showed an increase in amount of saliva in response to the food stimulus and eleven (p < 0·029, binomial test) had increased insulin values. For glucose and glucagon changes, subjects showed no statistically significant direction of change.
(N=4)
100
1\
,
I,
95
~
~
I 90f
I
~(N=14)
I
85,
W~O'"
80[1
i
1No,,) \ 1NO"/,\
~, (N=14~
75
-10
\
(N=14)~
_S--~20,-+-'c2-~+--:6c--~+~IO
t
Food presentation Time (mini
FIGURE I. Plasma insulin concentrations (means) at different times before and after the presentation of food.
213
SAllY AR Y AND INSULIN RESPONSES TO FOOD
TABLE 3 The number of subjects showing an increase, decrease, or no change in saliva, insulin, glucose or glucagon in response to the presentation of food
Increase Post> Pre. No change Post= Pre. Decrease Post
Saliva
Insulin
Glucose
Glucagon
13
11
6
4
o
o
o
3
7
10
14
14
14
14
p<0'001
p<0'029
NS
NS
TABLE
4
Correlations between increases in salivation, body weight indices, and scores on the Herman and Polivy Scale Herman and Pol ivy Scale scores Increase in salivation W/H 2
rs= +0·60 (p
r,=+O'SS (p
* One-tailed test. There were statistically significant correlations between the degree of increase in salivation (mean pre-food stimuli values subtracted from mean post-food stimuli values), the body mass index and the Herman and Polivy Scale (Table 4). The degree of increase in insulin scores (mean pre-food stimuli values subtracted from mean postfood stimuli values) did not correlate significantly with the body mass index, the Herman and Polivy Scale or the salivary response. Neither the scores for food appeal nor scores on the Stunkard hunger rating scale correlated with the increase in salivation, in insulin or with any of the other measures. Seven of the ten questions on the Herman and Polivy Scale are also on the Stunkard and Messick Scale, thus allowing a test-retest correlation to be performed on those seven questions. The correlation between the first taking of the test and the second taking ofthe test, although highly significant (rs = + 0'87, p <0'01), became even greater with the removal of one subject's scores, those of subject 4 (rs= +0'94, p
214
B. 1. SAHAKIAN ET AL DISCUSSION
One of the main findings of this study was that non-obese men and women respond to the sight and smell of food by both salivating and increasing the insulin concentrations in peripheral blood. The salivary response has been documented in many studies by Wooley and Wooley (in press) who have used it as an index of involuntary appetite and observed increases in salivary output in response to food deprivation. The period of fasting in the current study was standardized to end at lunch-time after a 17 h fast so that the individual variability in preferences for eating at breakfast-time would not affect the responses and so that the standard food stimulus, previously tested for acceptability for all the subjects, would be considered appropriate for the time of day. This standardization may have helped to reveal the statistically significant relationships between the salivary response and both the subject's response rating in the Herman and Polivy questionnaire and his or her body mass index. The relationship of the salivary response to the body mass index would seem surprising given such a narrow weight range among the subjects, but this correlate was not distinguishable from the restraint scores since there was also a strong correlation between the body mass index and the degree of restraint. Five of the ten questions on the Herman and Pol ivy Scale are linked to awareness of weight changes. One explanation for these observations may then be that the relationship between salivation and body mass reflects predominantly a restraint effect rather than an innate link between the body's autonomic responsiveness, displayed by salivation and the body mass index. The alternative explanation that the salivary response in some way reflects the individual's ingestive behaviour and thereby the tendency to weight gain has been suggested (Wooley, Wooley & Woods, 1975), but is no more plausible an explanation for these findings than other suggestions. To distinguish between the effect of restraint and body weight would require the study of unrestrained individuals of widely different body weights and this type of study has not yet been performed. The findings in Table 2 of an insulin response to visual and olfactory cues from food is not new (Rodin, 1978), but does conflict with the careful studies of Sjostrom, Garellick, Krotkiewski and Luyckx (1980) who failed to find an insulin response in nonobese women. The change observed in the present study, as well as that from the study by Rodin (1978), is rather small. Careful triplicate analyses of plasma insulin are therefore necessary to display the effect and measurements need to undertaken 2-4 min after the stimulus has been applied if the effect is to be detected. Obese patients, with their insulin resistence and greater pancreatic secretion rates of insulin in response to a variety ofmetabolic or dietary stimuli, are more likely to have a detectable response to a neuronal effect mediated presumably via the vagus nerve. In the non-obese subject, therefore, great care is needed to ensure that the experimental procedures are undertaken with the minimum disturbance so that no adrenergic inhibition of insulin secretion is likely. Despite our efforts to minimize discomfort by the use of local anaesthesia for cannula insertion, two individuals in one group had to be eliminated from the analyses (Table 1). We were impressed by the need for careful organisation of blood taking and salivary collection in order to maintain an informal atmosphere, and consider this problem a potential error in studies of this nature. The other implication of the small plasma insulin response to food related stimuli is that this small increase in circulating insulin is unlikely in itself to mediate an increase in food intake in those with an enhanced response. Patients who are injected with
SALlY AR Y AND INSULIN RESPONSES TO FOOD
215
substantial amounts of insulin as part of insulin-hypo glycaemia tests rarely comment on hunger as a principal feature until the test is nearly over after 45-60 min and the blood glucose is beginning to rise. It seems unlikely therefore that the greater portal concentrations of insulin after neuronal activation could mediate an effect on hepatic receptors governing food intake. It seems more likely that insulin responses to food related stimuli, if eventually proven to correlate with eating behaviour, are not a primary link in the control of ingestive behaviour, but an index of behavioural responsiveness. The response did not, however, seem to correlate with analogue scores of hunger or of food acceptability.
CONCLUSION
To our knowledge, this is the first report of concurrent measurement of salivation and insulin secretion in response to food stimuli in the same subjects. The poor correlation within the group between the salivation and insulin responses suggests that these two responses, although affected by the autonomic system, need not be controlled by a common effector mechanism. We conclude that this system of testing for physiological correlates of appetite needs extending in a standardized fashion to subjects of different body weight with known responses to the current methods of assessing degree of eating restraint. There is also, however, a need to validate the use of these physiological tests to predict ingestive behaviour and then to explore whether the peripheral hormonal and gastrointestinal responses are among the mechanisms mediating an individual's eating pattern in response to visual and olfactory cues or merely an indirect manifestation of a central mechanisin. REFERENCES
Fuller. J. Human appetite and hodr si:=e control. the roles of indiridual differences and food dependencies in hU/11an appetite and hody size control processe,\. Unpublished doctoral dissertation, University of Birmingham, 1980. Polivy, J., Herman, C. P., & Warsh, S. Internal and external components of emotionality in restrained and unrestrained eaters. Journal of Abnormal Psychology, 1978,87,497-504. Powley, T. L. The ventromedial hypothalamic syndrome, satiety. and a cephalic phase hypothesis. Psychological Reriew, 1977,84,89-126. Rodin, J. Has the distinction between internal versus external control of feeding outlived its usefulness? In G. Bray (Ed .), Recent advances in obesity research (Vol. 2). Pp. 75-85. London: Newman, 1978. Runyon, R. P., & Haber, A. Fundamentals of behavioral statistics. Reading, Massachusetts: Addison-Wesley Publ. Co., 1967. Siegel, S. Non-parametric statistics for the behavioral sciences. London: McGraw-Hili, 1956. Silverstone, T. Drugs, appetite, and obesity in man. In G. Bray (Ed.), Recent advances in obesity research (Vol. 2). Pp.421-432. London: Newman, 1978. Sjostrom, L., Garellick, G., Krotkiewski, M., & Luyckx, A. Peripheral insulin in response to the . sight and smell of food. Metabolism, 1980,29,901-909. Stunkard, A. 1. "Restrained eating": what it is and a new scale to measure it. In L. A. Cioffi, W. P. T. James & T. Van Itallie (Eds.), Body weight regulatory systems: normal and disturbed mechanism. New York: Raven Press, 1981. Wooley, O. W., Wooley, S. c., & Woods, W.A. Effect of calories on appetite for palatable food in obese and non-obese humans. Journal of Comparative and Physiological Psychology, 1975, 89,619-625.
216
B. J. SAHAKIAN ET AL
Wooley. O. W .. & Wooley, S. C. Relationship of salivation in humans to deprivation. inhibition and the encephalization of hunger. Appetite: Journalfor Research Intake, 2, in press. Wooley, S. C., & Wooley, O. W. Salivation to the sight and thought of food: a new measure of appetite. Ps\'chosomatic Medicine, 1973,35, 136-142.
Received 1 June, 1981