www.elsevier.com/locate/ynimg NeuroImage 24 (2005) 363 – 368 Functional MRI of human hypothalamic responses following glucose ingestion Paul A.M. Smeets,a,b,* Cees de Graaf,b Annette Stafleu,b Matthias J.P. van Osch,a and Jeroen van der Gronda a Image Sciences Institute, University Medical Center Utrecht, Utrecht, The Netherlands TNO Nutrition and Food Research, Zeist, The Netherlands b Received 19 March 2004; revised 1 July 2004; accepted 16 July 2004 Available online 13 November 2004 The hypothalamus is intimately involved in the regulation of food intake, integrating multiple neural and hormonal signals. Several hypothalamic nuclei contain glucose-sensitive neurons, which play a crucial role in energy homeostasis. Although a few functional magnetic resonance imaging (fMRI) studies have indicated that glucose consumption has some effect on the neuronal activity levels in the hypothalamus, this matter has not been investigated extensively yet. For instance, dose-dependency of the hypothalamic responses to glucose ingestion has not been addressed. We measured the effects of two different glucose loads on neuronal activity levels in the human hypothalamus using fMRI. After an overnight fast, the hypothalamus of 15 normal weight men was scanned continuously for 37 min. After 7 min, subjects ingested either water or a glucose solution containing 25 or 75 g of glucose. We observed a prolonged decrease of the fMRI signal in the hypothalamus, which started shortly after subjects began drinking the glucose solution and lasted for at least 30 min. Moreover, the observed response was dose-dependent: a larger glucose load resulted in a larger signal decrease. This effect was most pronounced in the upper anterior hypothalamus. In the upper posterior hypothalamus, the signal decrease was similar for both glucose loads. No effect was found in the lower hypothalamus. We suggest a possible relation between the observed hypothalamic response and changes in the blood insulin concentration. D 2004 Elsevier Inc. All rights reserved. Keywords: Hypothalamus; Blood; Glucose; BOLD fMRI Introduction An important part of the brain involved in the regulation of food intake is the hypothalamus. Many of its nuclei contain glucose* Corresponding author. University Medical Center Utrecht, Heidelberglaan 100, Room E01.335, 3584 CX Utrecht, The Netherlands. Fax: +31 30 251 3399. E-mail address: paul@isi.uu.nl (P.A.M. Smeets). Available online on ScienceDirect (www.sciencedirect.com). 1053-8119/$ - see front matter D 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.neuroimage.2004.07.073 sensitive neurons, which are thought to play an important role in long-term body weight regulation as well as in acute feeding behavior (Oomura, 1980; Williams et al., 2001). In recent years, functional magnetic resonance imaging (fMRI) has provided an indirect but non-invasive way to measure changes in neuronal activity levels in the brain of awake subjects. Rather than measuring neuronal spiking activity, this technique measures a vascular correlate of neuronal activity, the Blood Oxygen LevelDependent signal (BOLD signal). The BOLD signal changes due to the changes in the local concentrations of oxygenated and deoxygenated hemoglobin that result from the hemodynamic changes associated with neuronal activity (Attwell and Iadecola, 2002; Ogawa et al., 1992). The small size of the hypothalamus, its nuclei and the small signal changes in fMRI make it difficult and technically demanding to image the effects of food stimuli in this part of the brain. Still, a few fMRI studies have shown effects of glucose administration on neuronal activity levels in the hypothalamus. Transient changes of the BOLD signal in the hypothalamus after administration of glucose have been reported in both rats (Mahankali et al., 2000; Torii et al., 1997) and humans (Gao et al., 1998; Liu et al., 2000; Matsuda et al., 1999). Although the signal changes reported in these studies are not entirely consistent with each other, it is clear that glucose administration somehow results in a response in the hypothalamus. Likely, the rise in blood glucose concentration that follows the administration of glucose is an important factor mediating this response. These studies have demonstrated that it is feasible to measure long-term effects of a food stimulus in the human hypothalamus with BOLD fMRI. However, to enable the use of this method in future studies comparing the effects of food or drug stimuli in different groups of subjects, for example, comparing the effects of a glucose stimulus in obese and normalweight subjects, reproducibility of the method and dose-dependency of the measured response need to be addressed. Therefore, the purpose of our study was twofold: first, to try and replicate the kind of BOLD measurements performed in earlier studies to examine the temporal profile of the hypothalamic response to a glucose load 364 P.A.M. Smeets et al. / NeuroImage 24 (2005) 363–368 and second, to investigate whether the amount of glucose ingested affects the hypothalamic response. aligned to the middle image. The anatomical image was also coregistered with this image. Data analysis Materials and methods Subjects Fifteen healthy normal weight male volunteers participated, mean age 21.9 (SD 3.1) years, BMI 21.5 (SD 1.9) kg/m2. Subjects were recruited by an advertisement put up at various locations in the University Medical Center Utrecht. We used a Health and Lifestyle Questionnaire to assess general health and aspects of lifestyle relevant to the study. Exclusion criteria included: having a body mass index (BMI) lower than 19 kg/m2 or higher than 25 kg/m2; being under 18 or over 28 years of age at the study day; smoking; having a history of alcohol consumption or current alcohol consumption of more than 28 units per week; having a history of medical or surgical events that may significantly affect the study outcome, such as metabolic or endocrine disease or any gastrointestinal disorder; having irregular eating habits; slimming or following a medically prescribed diet; using medication (except aspirin/paracetamol); suffering from claustrophobia; having diabetes; having metal implants or metal objects on the body which cannot be removed (e.g., piercing, hearing aid, brace). Written informed consent was obtained from all subjects according to the Declaration of Helsinki and the study protocol was approved by the Medical Ethical Committee of the University Medical Center Utrecht, Utrecht, The Netherlands. Every subject’s hypothalamus was manually segmented using the anatomical image and divided into four regions of interest by two orthogonal axes following predefined criteria (Matsuda et al., 1999): the upper anterior hypothalamus (UAH), lower anterior hypothalamus (LAH), upper posterior hypothalamus (UPH), and lower posterior hypothalamus (LPH). The anterior–posterior axis was defined by the line passing through the centers of the anterior commissure and the mammillary body. The upper–lower axis was determined by the line passing through the optic chiasm, perpendicular to the anterior–posterior axis (Fig. 1). Also, a square reference area (10  10 pixels) of about the same size as the hypothalamus was delineated in the frontal cortex, anterior of the genu of the corpus callosum. At every time point, the mean gray value in the hypothalamus as a whole and in each region of interest was calculated. Next, these mean gray values were normalized to the mean of the 7.2-min baseline, yielding the percentage signal change from the mean baseline. For statistical analysis we pooled the normalized data of every group in time, resulting in 37 time slots of 1 min. Hereafter, we tested for every time slot whether the two glucose conditions and the water condition differed. Statistical testing was performed by a Student’s t test with a Bonferroni corrected threshold of P = 0.0001. This method is comparable to differential regression analysis (Cho et al., 2003). Results Experimental procedures Subjects were instructed to fast overnight from 10 pm until the scan the next morning, which started between 9:30 and 10:00 am (no food or beverages, except water). They were randomly assigned to one of three experimental conditions: 300 ml of orange-flavored water in which 25 or 75 g of d-dextrose (Avebe Corporate, Veendam, The Netherlands) was dissolved (n= 5 and n= 6, respectively), or 300 ml plain tap water (n= 4). Magnetic resonance imaging was performed using a 1.5-T Philips Gyroscan ACS-NT system. Subjects were positioned supine with their head immobilized by a vacuum cushion designed for use in a MRI headcoil (Medical Intelligence, Schwabmqnchen, Germany). During the functional scan, a 10-mm thick midsagittal slice was scanned using a T2*-weighted gradient-echo segmented echo-planar imaging (EPI) sequence (TR = 120 ms, TE = 40 ms, flip angle = 308, image matrix = 198  256, FOV = 208  208 mm, 12 signal averages per scan, 33 k-lines acquired per excitation pulse, adapted from Liu et al. (2000)). Images were reconstructed to 256  256 pixels. Subjects were scanned for 37 min (256 scans). After a baseline of 7.2 min (50 scans), subjects ingested one of the test solutions through a peroral tube. After the functional scan, a T1-weighted anatomical scan was made of the same slice (TR = 600 ms, TE = 18 ms, FOV = 230  230 mm). Data preprocessing All 256 functional images of each time series were motion corrected with in-house software, which uses the MIRIT mutual information registration routine (Maes et al., 1997). Images were The mean signal changes in the hypothalamus and the reference area, as a function of time, are shown in Fig. 2A. At the start of drinking (t = 0 min), large signal drops occur for all treatments, in both the hypothalamus and the reference area. These result from artifacts caused by the drinking and last for about 3 min, obscuring possible fMRI signal changes. After that, both glucose treatments show a prolonged signal decrease (1–2.5%), whereas the water treatment returns to baseline. Moreover, the 75-g glucose solution induces a larger decrease in signal than the 25-g glucose solution. Fig. 1. Segmentation and subdivision of the hypothalamus into four regions of interest (Matsuda et al., 1999). UA: upper anterior hypothalamus; UP: upper posterior hypothalamus; LA: lower anterior hypothalamus; LP: lower posterior hypothalamus; ac: anterior commissure; mm: mammillary body; oc: optic chiasm. P.A.M. Smeets et al. / NeuroImage 24 (2005) 363–368 365 Fig. 2. (A) Mean signal change from the mean baseline in time for the hypothalamus as a whole and for a reference area of comparable size under three conditions. (B) P values of the Student’s t tests comparing the mean signal changes from the mean baseline of the two glucose conditions with that of the water condition for every 1 min time slot. The dashed line indicates the Bonferroni corrected threshold of P = 0.0001. Time t = 0 min corresponds to the onset of drinking and the black bar indicates the approximate duration of drinking. Fig. 2B shows that, in the hypothalamus, the signal after ingestion of both glucose solutions is significantly lower than that after the ingestion of water, and that this effect persists for an extended period of time. In the 25-g glucose condition, the signal differs significantly from the water condition until about 25 min after the onset of drinking. In the 75-g glucose condition, the signal differs significantly from the water condition for the duration of the scan (30 min). In the reference area, there are no significant signal changes after drinking for any of the treatments until the end of the scan (t = 25 min), where the difference between the water and the 75-g glucose solution becomes significant (Fig. 2B). In Fig. 2A, it can be seen that this is due to a slight decline in the signal of the water condition in the reference area, which is also apparent for the 25-g glucose condition. We attribute this drop in signal to scanner signal drift and consider it unrelated to the glucose content of the stimuli employed since both the water and the 25-g glucose condition show a similar signal decline. In addition to the hypothalamus as a whole, four sub-regions were studied (see Fig. 1). Fig. 3A shows the mean signal change as a function of time in these regions for the three conditions. The decrease in signal after glucose ingestion is present in the upper anterior hypothalamus (UAH) as well as in the upper posterior hypothalamus (UPH). In the UAH, this signal decrease is dosedependent: around 1% for the 25-g glucose condition and 2–3% for the 75-g glucose condition. In the UPH, both treatments show a similar signal decrease of 1–2%. In the lower posterior hypothalamus (LPH), a small, but mostly not significant, signal decrease can be seen for the 75-g glucose condition. In the lower anterior hypothalamus (LAH), there are no significant signal changes for both glucose treatments. The water condition shows small signal increases (0.5–1%) in the UPH, AUH and LPH, but not in the LAH and the reference area. Discussion The most important findings of this study are twofold. First, we observed a prolonged decrease in the fMRI signal in the hypothalamus following the ingestion of glucose. Second, this decrease was dose-dependent. We are the first to report a prolonged decrease in the hypothalamic fMRI signal after glucose consumption. The time course of this response suggests that it is associated with the changes in blood insulin and glucose concentration. Although the exact onset of this signal decrease cannot be determined accurately due to image artefacts associated with drinking, it starts before the end of glucose ingestion. This is well before most of the glucose has entered the blood stream. Therefore, initially, the observed response cannot be solely associated with the rise in blood glucose. Possibly, it is the brain that triggers anticipatory changes in blood insulin, in response to the consumption of a glucose solution. Rolls et al. (1976) showed that neurons in the lateral hypothalamic area of hungry monkeys decreased their firing rate at the sight of food, which shows that the hypothalamus can respond to a food stimulus, even before the actual onset of feeding. Early insulin secretion (0– 10 min after the onset of feeding), also called cephalic phase insulin release (CPIR), has been the subject of many studies (Bellisle et al., 1985; Berthoud et al., 1981; Lucas et al., 1987; Teff and Engelman, 1996; Teff et al., 1991, 1993). Although a relatively small amount of insulin is secreted, it is important in preventing post-prandial peaks in blood glucose (Ahren and Holst, 2001; Harju and Nordback, 1987; Kraegen et al., 1981). Some authors report the absence of CPIR in humans in response to tasting, but not actually swallowing, sweetened liquids (Bruce et al., 1987; Teff et al., 1995). Still, others have shown that the sight and sweet taste of food can trigger CPIR (Rodin, 1985; Yamazaki and Sakaguchi, 366 P.A.M. Smeets et al. / NeuroImage 24 (2005) 363–368 Fig. 3. (A) Mean signal change from the mean baseline in time for four subregions of the hypothalamus under three conditions. (B) P values of the Student’s t tests comparing the mean signal changes from the mean baseline of the two glucose conditions with that of the water condition for every 1 min time slot. The dashed line indicates the Bonferroni corrected threshold of P = 0.0001. Time t = 0 min corresponds to the onset of drinking and the black bar indicates the approximate duration of drinking. 1986). Moreover, several authors report that CPIR is mediated by the vagus nerve (Ahren and Holst, 2001; Harju and Nordback, 1987; Yamazaki and Sakaguchi, 1986), which projects to the hypothalamus (Oomura, 1980; Storlien, 1985). In our experiment, subjects swallowed sweet colored glucose solutions which, however, they could not see during the scan. Apparently, swallowing 300 ml of a sweet solution when hungry was sufficient to trigger CPIR, or at least an anticipatory hypothalamic response, in our subjects. After CPIR, normal post-prandial insulin secretion comes into play when blood glucose starts rising. Blood glucose peaks at about 30 min after ingestion of 75 g glucose in a standard 75 g oral glucose tolerance test (Kong et al., 1999; Yasuhara et al., 2003). The insulin concentration changes concomitantly (Kong et al., 1999). Our observation of a dose-dependent modulation of the fMRI signal in the hypothalamus, where a larger dose of glucose was associated with a larger and more prolonged signal decrease, fits the pattern of glucose-triggered insulin release. P.A.M. Smeets et al. / NeuroImage 24 (2005) 363–368 In our design, the effects of taste and energy content of the stimuli on the hypothalamic response cannot be separated. In future research, taste effects could be ruled out by injecting solutions directly into the stomach, which would also bypass the problem of swallowing artifacts. The effect of glucose administration on the hypothalamus has been studied previously in rats and humans (Liu et al., 2000; Mahankali et al., 2000; Matsuda et al., 1999; Torii et al., 1997). Matsuda et al. (1999) reported a large signal decrease (8–10%) in the hypothalamus, starting 4 min after the onset of drinking and lasting about 10 min, whereas Liu et al. (2000) found a signal decrease of up to 4%, starting 5 min after the onset of drinking, reaching a maximum around 8 min and returning to about 1% below baseline after 12 min. We found no pronounced effect of water intake. This corresponds with the finding of Liu et al. (2000), but contrasts with the finding of Matsuda et al. (1999) who report a signal decrease in part of the hypothalamus after the ingestion of water similar to that following glucose ingestion. The decreases in fMRI signal we observe in the hypothalamus are possibly related to decreases in neuronal activity in the lateral hypothalamic area (LHA), which is known to contain glucosesensitive neurons (Oomura, 1980). Moreover, decreased firing rates in response to glucose infusion have been reported in this area in cats and rats (Brown and Melzack, 1969; Chhina et al., 1971; Miller and Rabin, 1975; Oomura et al., 1974). We found a prolonged decrease of the fMRI signal in the hypothalamus. An important issue in the interpretation of this result is localization. The signal in fMRI represents local changes in blood oxygenation. This signal is obtained from several voxels (volume units of brain tissue), whose size sets the spatial scanning resolution. In our case, the voxels are 1  1  10 mm. It is important to realize, however, that what we measure in fMRI is not the spiking activity of the multiple neurons present in a voxel, but rather the local changes in blood oxygenation and blood flow caused by a changing level of neural activity (Attwell and Iadecola, 2002). Because the fMRI signal relies on the hemodynamic changes associated with changes in neuronal activity, it does not co-localize perfectly with the neurons involved. Thus, localization of fMRI responses is not as accurate as that of the single cell or multi-unit electrical recordings made in animals, which have shown the modulation of neuronal activity in the ventromedial and lateral hypothalamus in response to glucose (Brown and Melzack, 1969; Chhina et al., 1971; Miller and Rabin, 1975; Oomura et al., 1969; Rabin and Miller, 1980). We imaged a 10-mm midsagittal slice which includes all of the hypothalamus in the anterior–posterior direction and most, if not all of it, in the medio-lateral direction (Saper, 1990). This is the preferred orientation in this case because most movement associated with drinking comes from in-plane rotation and preventing side-ways motion of the head is relatively easy. The decrease in fMRI signal, as observed in the hypothalamus as a whole, is not present in all subdivisions (see Fig. 1). Thus, this response associated with glucose ingestion is present only in part of the hypothalamus. It is tempting to try and identify one of the hypothalamic nuclei as the hot spot of signal change. However, this should be done with extreme caution for two reasons: First, because the fMRI signal does not directly, but indirectly, represent responses of groups of neurons and second, because every voxel contains parts of more than one nucleus. For example, in the ventral mid-tuberal region a typical voxel will, 367 apart from 3rd ventricle CSF, contain cells belonging to both the ventro-medial nucleus (about 2 mm in cross-section) and the adjacent lateral hypothalamus (about 3 mm in cross-section) (Saper, 1990). Regardless of its exact localization, the dose-dependent decrease we found in the UAH might provide a measure of satiation if this decrease relates to changes in the blood insulin concentration. Recently, it has been shown that the time-course of the satiating effect of carbohydrates relates to their glycemic effect. High-glycemic carbohydrates, which cause a rapid rise of the blood glucose concentration, increase satiety and suppress food intake mostly in the short term (within 1 h), whereas low-glycemic carbohydrates, which cause a more gradual change in blood glucose concentration, increase satiety, and suppress food intake on the longer term (6 h) (Anderson and Woodend, 2003; Anderson et al., 2002). The response we observed might be associated with this since it is associated with the glycemic changes induced by glucose ingestion. This, however, will require a more in-depth investigation of the correlation between the changes in hypothalamic fMRI signal, blood glucose and insulin concentrations, and the degree of satiation. In conclusion, this is the first study showing a prolonged and dose-dependent decrease of the fMRI signal in the hypothalamus after glucose ingestion. The dose-dependency of the signal decrease was exclusively present in one subdivision of the hypothalamus: the upper anterior hypothalamus. The time course and dose-dependency of this response suggest a possible association with changes in the blood insulin concentration. This, however, will require further research. Acknowledgments This study was financially supported by the Dutch Ministry of Education, Culture and Science and the Dutch Ministry of Health, Welfare and Sport. References Ahren, B., Holst, J.J., 2001. The cephalic insulin response to meal ingestion in humans is dependent on both cholinergic and noncholinergic mechanisms and is important for postprandial glycemia. Diabetes 50, 1030 – 1038. Anderson, G.H., Woodend, D., 2003. Effect of glycemic carbohydrates on short-term satiety and food intake. Nutr. Rev. 61, S17 – S26. Anderson, G.H., Catherine, N.L., Woodend, D.M., Wolever, T.M., 2002. Inverse association between the effect of carbohydrates on blood glucose and subsequent short-term food intake in young men. Am. J. Clin. Nutr. 76, 1023 – 1030. Attwell, D., Iadecola, C., 2002. The neural basis of functional brain imaging signals. Trends Neurosci. 25, 621 – 625. Bellisle, F., Louis-Sylvestre, J., Demozay, F., Blazy, D., Le Magnen, J., 1985. Cephalic phase of insulin secretion and food stimulation in humans: a new perspective. Am. J. Physiol. 249, E639 – E645. Berthoud, H.R., Bereiter, D.A., Trimble, E.R., Siegel, E.G., Jeanrenaud, B., 1981. Cephalic phase, reflex insulin secretion. Neuroanatomical and physiological characterization. Diabetologia 20, 393 – 401. Brown, K.A., Melzack, R., 1969. Effects of glucose on multi-unit activity in the hypothalamus. Exp. Neurol. 24, 363 – 373. Bruce, D.G., Storlien, L.H., Furler, S.M., Chisholm, D.J., 1987. Cephalic phase metabolic responses in normal weight adults. Metabolism 36, 721 – 725. 368 P.A.M. Smeets et al. / NeuroImage 24 (2005) 363–368 Chhina, G.S., Anand, B.K., Singh, B., Rao, P.S., 1971. Effect of glucose on hypothalamic feeding centers in differentiated animals. Am. J. Physiol. 221, 662 – 667. Cho, Z.H., Son, Y.D., Kang, C.K., Han, J.Y., Wong, E.K., Bai, S.J., 2003. Pain dynamics observed by functional magnetic resonance imaging: differential regression analysis technique. J. Magn Reson. Imaging 18, 273 – 283. Gao, J.H., Fox, P.T., Liu, Y.J., Pu, Y., DeFronzo, R.A., Matsuda, M., 1998. fMRI of the hypothalamic region in subjects with obesity. NeuroImage 8, S926. (Suppl.). Harju, E., Nordback, I., 1987. Effect of atropine on insulin secretion in healthy subjects. J. Int. Med. Res. 15, 167 – 169. Kong, M.F., Chapman, I., Goble, E., Wishart, J., Wittert, G., Morris, H., Horowitz, M., 1999. Effects of oral fructose and glucose on plasma GLP-1 and appetite in normal subjects. Peptides 20, 545 – 551. Kraegen, E.W., Chisholm, D.J., McNamara, M.E., 1981. Timing of insulin delivery with meals. Horm. Metab. Res. 13, 365 – 367. Liu, Y., Gao, J.H., Liu, H.L., Fox, P.T., 2000. The temporal response of the brain after eating revealed by functional MRI. Nature 405, 1058 – 1062. Lucas, F., Bellisle, F., Di Maio, A., 1987. Spontaneous insulin fluctuations and the preabsorptive insulin response to food ingestion in humans. Physiol. Behav. 40, 631 – 636. Maes, F., Collignon, A., Vandermeulen, D., Marchal, G., Suetens, P., 1997. Multimodality image registration by maximization of mutual information. IEEE Trans. Med. Imaging 16, 187 – 198. Mahankali, S., Liu, Y., Pu, Y., Wang, J., Chen, C.W., Fox, P.T., Gao, J.H., 2000. In vivo fMRI demonstration of hypothalamic function following intraperitoneal glucose administration in a rat model. Magn. Reson. Med. 43, 155 – 159. Matsuda, M., Liu, Y., Mahankali, S., Pu, Y., Mahankali, A., Wang, J., DeFronzo, R.A., Fox, P.T., Gao, J.H., 1999. Altered hypothalamic function in response to glucose ingestion in obese humans. Diabetes 48, 1801 – 1806. Miller, D.S., Rabin, B.M., 1975. Effects of glucose on multiple-unit activity of the hypothalamus in female rats. Exp. Neurol. 49, 418 – 428. Ogawa, S., Tank, D.W., Menon, R., Ellermann, J.M., Kim, S.G., Merkle, H., Ugurbil, K., 1992. Intrinsic signal changes accompanying sensory stimulation: functional brain mapping with magnetic resonance imaging. Proc. Natl. Acad. Sci. U. S. A. 89, 5951 – 5955. Oomura, Y., 1980. Input–output organization in the hypothalamus relating to food intake behavior. In: Morgane, P.J., Panksepp, J. (Eds.), Physiology of the Hypothalamus. Marcel Dekker, New York, pp. 557 – 620. Oomura, Y., Ono, T., Ooyama, H., Wayner, M.J., 1969. Glucose and osmosensitive neurones of the rat hypothalamus. Nature 222, 282 – 284. Oomura, Y., Ooyama, H., Sugimori, M., Nakamura, T., Yamada, Y., 1974. Glucose inhibition of the glucose-sensitive neurone in the rat lateral hypothalamus. Nature 247, 284 – 286. Rabin, B.M., Miller, D.S., 1980. Effects of 2-deoxy-d-glucose on multiunit activity of the hypothalamus in rats. J. Neurosci. Res. 5, 233 – 239. Rodin, J., 1985. Insulin levels, hunger, and food intake: an example of feedback loops in body weight regulation. Health Psychol. 4, 1 – 24. Rolls, E.T., Burton, M.J., Mora, F., 1976. Hypothalamic neuronal responses associated with the sight of food. Brain Res. 111, 53 – 66. Saper, C.B., 1990. Hypothalamus. In: Paxinos, G. (Ed.), The Human Nervous System. Academic Press, San Diego, pp. 389 – 413. Storlien, L.H., 1985. The ventromedial hypothalamic area and the vagus are neural substrates for anticipatory insulin release. J. Auton. Nerv. Syst. 13, 303 – 310. Teff, K.L., Engelman, K., 1996. Oral sensory stimulation improves glucose tolerance in humans: effects on insulin, C-peptide, and glucagon. Am. J. Physiol. 270, R1371 – R1379. Teff, K.L., Mattes, R.D., Engelman, K., 1991. Cephalic phase insulin release in normal weight males: verification and reliability. Am. J. Physiol. 261, E430 – E436. Teff, K.L., Levin, B.E., Engelman, K., 1993. Oral sensory stimulation in men: effects on insulin, C-peptide, and catecholamines. Am. J. Physiol. 265, R1223 – R1230. Teff, K.L., Devine, J., Engelman, K., 1995. Sweet taste: effect on cephalic phase insulin release in men. Physiol. Behav. 57, 1089 – 1095. Torii, K., Kondoh, T., Mori, M., 1997. Functional MRI changes in the hypothalamus of streptozotocin-treated rats after glucose loading. Diabetologia 40 (Suppl. 1), A240. Williams, G., Bing, C., Cai, X.J., Harrold, J.A., King, P.J., Liu, X.H., 2001. The hypothalamus and the control of energy homeostasis: different circuits, different purposes. Physiol. Behav. 74, 683 – 701. Yamazaki, M., Sakaguchi, T., 1986. Effects of d-glucose anomers on sweetness taste and insulin release in man. Brain Res. Bull. 17, 271 – 274. Yasuhara, D., Naruo, T., Nagai, N., Tanaka, M., Muranaga, T., Nozoe, S., 2003. Insulinogenic index at 15 min as a marker of nutritional rehabilitation in anorexia nervosa. Am. J. Clin. Nutr. 77, 292 – 299.