Diabetes Volume 65, August 2016 2179

Haiko Schlögl,1 Karsten Müller,2 Annette Horstmann,2,3 Konstanze Miehle,1 Janett Püschel,1,3 Arno Villringer,2,4 Burkhard Pleger,2,5 Michael Stumvoll,1 and Mathias Fasshauer1,3

Leptin Substitution in Patients With : Neural Correlates for Long-term Success in the Normalization of Eating Behavior

Diabetes 2016;65:2179–2186 | DOI: 10.2337/db15-1550

Lipodystrophy (LD) is a rare disease with a paucity of of . Several genetic mutations are known to cause subcutaneous adipocytes and leptin deficiency. Patients partial or generalized forms of the disease. Also, cases of often develop severe diabetes and, additionally, show a acquired LD have occurred (1). LD is frequently accompa- disturbed eating behavior with reduced satiety. The dis- nied by type 2 diabetes and dyslipidemia. Leptin substi- turbed eating behavior can be restored by substitution tution in the form of the analog metreleptin has shown with the leptin analog metreleptin. Long-term effects of beneficial metabolic effects. In the biggest clinical trial on metreleptin on resting state brain connectivity in treatment- metreleptin treatment in LD so far, hemoglobin A1c (HbA1c) naive patients with LD have not been assessed. In this on average decreased by 1.5% and serum triglycerides (TGs) study, resting state functional MRI scans and extensive fell by .50% after 1 year of treatment (2). Additionally, a behavioral testing assessing changes in hunger/satiety disturbed eating behavior often develops in patients with regulation were performed during the first 52 weeks LD, with reduced satiety after food consumption, leading to of metreleptin treatment in nine patients with LD. Rest- an increase in meal frequency (3). Impaired eating behavior ing state connectivity significantly increased over the course of metreleptin treatment in three brain areas can be improved by leptin substitution, too (4). Humoral (i.e., hypothalamus, insula/superior temporal gyrus, me- leptin crosses the blood-brain barrier by active transport dial prefrontal cortex). Behavioral tests demonstrated in the proximity of the mediobasal hypothalamus where that perceived hunger, importance of eating, eating fre- the blood-brain barrier is well permeable for peripheral quencies, and liking ratings of food pictures signifi- hormones (3,5). Through receptors in the arcuate nucleus, cantly decreased during metreleptin therapy. Taken leptin inhibits food intake by direct activation of anorex- together, leptin substitution was accompanied by long- igenic cocaine- and amphetamine-regulated transcript term changes of hedonic and homeostatic central ner- and neurons (6,7) as well as by in- vous networks regulating eating behavior as well as hibition of orexigenic Y neurons (8). In a decreased hunger feelings and diminished incentive widely accepted model of the control of eating behavior, value of food. Future studies need to assess whether the hypothalamus is considered to govern the homeostatic metreleptin treatment in LD restores physiological component of human eating regulation, that is, the drive to processes important for the development of satiety. eat to meet the bodily demands for energy (9,10). In addition, the is also expressed in neurons of the mesolimbic dopamine system, which is Lipodystrophy (LD) is a rare disease with a paucity of sub- involved in the processing of motivation and reward. Here, cutaneousadipocytesandreduced blood concentrations leptin reduces dopamine signaling (11). These findings also

1Department of Medicine, University Hospital Leipzig, Leipzig, Germany Received 9 November 2015 and accepted 5 May 2016. 2 Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany This article contains Supplementary Data online at http://diabetes 3 Integrated Research and Treatment Center AdiposityDiseases, University of .diabetesjournals.org/lookup/suppl/doi:10.2337/db15-1550/-/DC1. Leipzig, Leipzig, Germany H.S. and K.Mü. contributed equally to this study. 4Day Clinic of Cognitive Neurology, University of Leipzig, Leipzig, Germany 5Department of Neurology, BG University Clinic Bergmannsheil, Ruhr University © 2016 by the American Diabetes Association. Readers may use this article as Bochum, Bochum, Germany long as the work is properly cited, the use is educational and not for profit, and the work is not altered. Corresponding author: Haiko Schlögl, [email protected]. 2180 Neural Effects of Leptin in Lipodystrophy Diabetes Volume 65, August 2016 suggest an influence of leptin on the other component Medication contributing to the formation of human eating behavior, The leptin analog metreleptin was used for treatment. that is, hedonic eating, which is the drive to eat for plea- Metreleptin was provided by (San Diego, CA), sure in the absence of an energy deficit (12). Human be- Bristol-Myers Squibb (Munich, Germany), AstraZeneca havioral data support the hypothesis that leptin affects (London, U.K.), and Aegerion Pharmaceuticals (Cambridge, both components of eating control (4,13). MA) and applied subcutaneously. Dosing was recom- Apart from the clinical need for metreleptin treatment, mended by the respective manufacturer to achieve phys- patients with decreased blood concentrations of leptin iological replacement. Patients 1–3 (all female) were provide a unique model to study the central nervous administered metreleptin 0.04 mg/kg body weight/day effects of leptin. However, in vivo investigations of b.i.d. for the first week and 0.08 mg/kg body weight/day metreleptin effects on brain activation assessed with thereafter, resulting in doses between 2.9 and 7.8 mg/day. functional MRI (fMRI) have not been performed in For patients 4–9, dosing changed due to modified instruc- treatment-naive patients with LD so far. Furthermore, tions by the manufacturer. Metreleptin was administered the impact of leptin on resting state brain function, which once daily at 2.5 mg/day for men and 5 mg/day for women, is a useful tool for the assessment of connectivity of brain independent of body weight. regions and long-term state changes (14), has not been Experimental Design investigated yet. To address these open points, the long- Experiments were performed between 2010 and 2014. term effects of leptin substitution on resting state brain Behavioral tests and MRI scanning were performed at six connectivity in treatment-naive patients with LD were time points: 1 day before start of metreleptin supple- assessed with task-free resting state fMRI before and at mentationandafter1,4,12,26,and52weeksof five time points over 52 weeks after initiation of metre- metreleptin treatment. On each measurement day, the leptin treatment. Because the anchoring of fMRI findings same set of questionnairesandbehavioraltestswas with behavioral results is crucial for a correct interpreta- completed as described in the next section. All patients tion (15), all patients underwent an extensive neuropsy- were asked to have a small lunch at 12:00 P.M. and to fast chological assessment accompanied by metabolic tests at thereafter until 5:00 P.M. Then, a standard meal consisting each study visit. of 20% of the daily energy requirements calculated for We hypothesized that with resting state fMRI, long- each patient was consumed. Because leptin physiologically term increases in connectivity in both homeostatic and mediates satiety, differences between leptin deficiency hedonicbrainareascanbeobserved and that these changes and imitated physiological leptin levels due to metreleptin are accompanied by behavioral changes, including perceived treatment were expected to be most pronounced in the hunger/satiety and liking ratings of food pictures. We sated state and, therefore, the mentioned calorie amount show for the first time combined long-term metreleptin waschosentocreateastateofmoderatesatiety.Wedid treatment–related effects on behavior and resting state not choose a higher percentage of daily energy require- brain connectivity in treatment-naive patients with leptin ments to avoid both a ceiling effect in extreme fullness deficiency. These findings are particularly important in view and postprandial tiredness during the following fMRI of recent successful approaches of overcoming leptin resis- scan. Before and after the meal, patients filled in vi- tance (16) and hopes of using metreleptin as a therapeutic sual analog scales (VASs). At 6:00 P.M., the MRI scan agent in obesity in the future. was performed. The next morning at 8:00 A.M.,afasting blood sample was taken for assessment of metabolic pa- RESEARCH DESIGN AND METHODS rameters, including fasting TG, HbA1c,andserumleptin Patients With LD concentrations. Nine patients with LD (seven female) eligible for metre- Questionnaires and Behavioral Tests leptin treatment at the University Hospital Leipzig par- Before all other measurements, the German versions of ticipated in the MRI study. Baseline characteristics and the Three Factor Eating Questionnaire (TFEQ) (17) (or laboratory data of included patients are summarized in Fragebogen zum Essverhalten [18]) and Inventory of Table 1. Inclusion criteria for leptin replacement were Eating Behavior and Weight Problems (Inventar zum established LD, age $5 years at baseline, insufficiently Essverhalten und Gewichtsproblemen [IEG]) (19) were com- controlled diabetes, and/or hypertriglyceridemia despite pleted. Because the TFEQ and IEG refer more to long-term adequate antihyperglycemic and -lowering medica- than to short-term attitudes and feelings, neither test was tion, respectively. Exclusion criteria included pregnancy performed 1 week after initiation of metreleptin treatment. or lactation, severe renal insufficiency, active malignant A complete list of the 14 scales of the IEG with German disease, primary hematologic abnormalities, infectious liver titles and English translations is shown as Supplementary disease, HIV infection, and hypersensitivity to Escherichia Table 1. The VAS were bars of 100 mm in length for the coli–derived proteins. All patients were metreleptin assessment of hunger and satiety feelings, where 0 mm in- treatment–naive and consented to participating in the dicated no hunger or satiety and 100 mm indicated extreme MRI study. hunger or satiety. The picture rating task was performed in diabetes.diabetesjournals.org Schlögl and Associates 2181

Table 1—Baseline characteristics and laboratory data of the study population

HbA1c Patient Phenotype of LD Mutation Sex Age (years) BMI (kg/m2) Leptin (mg/L) (% [mmol/mol]) TG (mmol/L) 1 Partial LMNA F 23 27.4 5.3 6.4 (46) 16.3 2 Partial LMNA F 34 28.7 5.2 8.9 (74) 2.4 3 Partial LMNA F 48 27.4 9.8 6.6 (49) 2.5 4 Generalized ND M 16 17.4 ,0.48 8.4 (68) 9.2 5 Partial ND F 41 30 3.6 8.0 (64) 7.9 6 Generalized ND F 33 20.7 ,0.48 8.1 (65) 37.1 7 Partial PPARg M 55 33.5 6.7 6.0 (42) 23.2 8 Partial ND F 39 30.8 11.2 7.6 (60) 15.6 9 Partial PPARg F 53 27.2 5.6 7.0 (53) 4.3 LMNA, lamin A/C; ND, not detected; PPARg, peroxisome proliferator–activated receptor-g. F, female; M, male.

a designated room with as little distraction as possible. To identify treatment-related connectivity changes, On a computer screen, 200 food and 50 nonfood pictures eigenvector centrality (EC) mapping was performed using were presented in random order and rated through a the Leipzig Image Processing and Statistical Inference keypad (four keys) for valence (food pictures: “How tasty Algorithms software package (21). EC provides a measure do you find the depicted food item?” [1 = not at all tasty; for detecting central hubs within a brain network by using 4 = extremely tasty]; nonfood pictures: “How much do an algorithm similar to the Google PageRank algorithm (22). you like the depicted object?” [1 = not at all; 4 = very For all voxels, a similarity matrix was generated, including much]). Pearson correlation coefficients, among all resting state fMRI Paradigm, Technical Parameters, and Data fMRI time courses. To use a similarity matrix with only Processing positive numbers, all negative entries were set to zero before The fMRI scan was performed in resting state. Patients computing the EC. In a second approach, we also used the inside the scanner were in a supine position and looking absolute value, taking all values of the similarity matrix into at a black screen with a white fixation cross. A whole-body accountwhencomputingtheECmeasure.Notethataccord- 3-T MAGNETOM Trio, A Tim System, scanner (Siemens, ing to the theorem of Peron and Frobenius (23), the simi- Erlangen, Germany) with a 32-channel head coil was larity matrix has a unique real largest eigenvalue, and the used. In each scanning session, resting state fMRI data corresponding eigenvector has strictly positive components. were acquired by using a gradient echo, echo planar im- Then, the EC map was generated by using the components aging sequence. The following parameters were used: 300 of this eigenvector to determine the EC of all voxels. whole-brain volumes, 64 3 64 acquisition matrix, and Within SPM, group-level statistical analysis of all six slice thickness of 4 mm (1-mm gap), resulting in a nom- EC maps for all subjects used the general linear model fl inal voxel size of 3 3 3 3 5mm3. Further imaging pa- with a exible factorial design with factors subject and rameters were 30 axial slices, repetition time of 2,300 ms, time. A weighted sum of the parameter estimates was echo time of 30 ms, flip angle of 90°, and bandwidth of statistically assessed by using a contrast vector generated 1,817 Hz/pixel. by the a priori hypothesis of an increased EC over time. Preprocessing of the fMRI data was performed by The resulting statistical parametric map was processed by P , using Statistical Parametric Mapping (SPM) 8 software, using a voxelwise threshold of 0.005. Taking the including estimation and correction for motion and multiple comparison problem into account, clusters were echo planar imaging deformation. The normalization was detected with a minimum size of 80 voxels to obtain clus- – P , performed by registering the individual three-dimensional ters with a false discovery rate (FDR) corrected 0.05. high-resolution T1-weighted structural image onto the functional images. This individual anatomical image was RESULTS further processed by the unified segmentation algorithm Anthropometric and Metabolic Parameters (20), and the resulting deformation field was applied onto in Patients With LD the functional images. After normalization, the resulting Baseline characteristics of all patients are summarized in voxel size of the functional images was interpolated to an Table 1. Mean 6 SEM age was 38 6 4 years, and BMI was isotropic voxel size of 3 3 3 3 3mm3. In the final step of 27.0 6 1.7 kg/m2. Baseline leptin was 5.3 6 1.2 mg/L, the preprocessing, the functional images were smoothed HbA1c was 7.4 6 0.3%, and TGs were 13.2 6 3.8 mmol/L. by using a Gaussian smoothing kernel of 8 mm full width Changes in BMI, HbA1c, and TGs during metreleptin at half maximum. treatment are summarized in Supplementary Table 2. 2182 Neural Effects of Leptin in Lipodystrophy Diabetes Volume 65, August 2016

Valence of Food/Nonfood Pictures and Hunger/Satiety remained significantly decreased compared with baseline Ratings up to 26 weeks (P , 0.05), and a strong trend was seen The average rating score of food pictures was 2.76 6 0.12 after 52 weeks (4.3 6 0.8, P = 0.052). The average base- and for nonfood pictures, 2.55 6 0.08 at baseline (Fig. line score for scale 3 (hunger) was 8.9 6 0.9, which also 1A). Valence of food pictures decreased after initiation decreased to 4.4 6 0.9 after 4 weeks of metreleptin treat- of metreleptin treatment, with significant decreases ob- ment (P , 0.01). The average score further decreased, served at 1 week, 4 weeks, and 52 weeks (P , 0.05) as with the lowest values detectable at 26 weeks (2.7 6 well as trends at 12 weeks (P = 0.06) and 26 weeks (P = 0.8, P , 0.001) and a significantly lower score detectable 0.07). In contrast, rating scores of nonfood pictures did after 52 weeks (3.1 6 1.1, P , 0.001). In contrast, aver- not change significantly throughout the 52 weeks of age scores for scale 1 (cognitive restraint of eating) did metreleptin treatment (Fig. 1A). not significantly change throughout the 52 weeks of metre- Fasting hunger rated on VAS was 54 6 9 mm at base- leptin treatment (Fig. 2A). line (Fig. 1B). Hunger ratings continuously decreased over On the IEG questionnaire, average scores for scale 1 the first 26 weeks of metreleptin treatment, with the (importance of eating), scale 2 (strength and triggering of lowest scores detectable at 26 weeks (18 6 6 mm, P = desire to eat), scale 9 (attitude towards obese persons), 0.002). Fasting hunger at 52 weeks was significantly and scale 11 (eating between meals) all significantly de- higher than at 26 weeks (44 6 9 mm, P = 0.04). Satiety creased at least at one time point after initiation of metre- rated 5 min after the meal was 72 6 8 mm at baseline. leptin treatment compared with baseline levels (Fig. 2B). Satiety ratings 5 min after the meal continuously in- Furthermore, the value for scale 7 (cognitive restraint of creased over the first 26 weeks of metreleptin treatment, eating) was significantly higher at 52 weeks of metreleptin with the highest scores detectable at 26 weeks (93 6 treatment (11.8 6 2.1) compared with baseline (9.0 6 2 mm, P = 0.03). Satiety rated 5 min after the meal at 1.9, P = 0.02) (Fig. 2B). In contrast, values on all other 52 weeks (72 6 8 mm) was significantly lower than at 26 scales were not significantly affected by metreleptin treat- weeks (P = 0.04) and equal to baseline levels. Satiety rated ment (data not shown). For original German wording 120 min after the meal was 56 6 9 mm at baseline and and English translation of scale titles, refer to Supple- increased to 78 6 5 mm after 1 week (P = 0.03) and 77 6 mentary Table 1. 8 mm after 4 weeks (P = 0.051). Satiety ratings 120 min fMRI Data after the meal at 12, 26, and 52 weeks were not signifi- Using resting state fMRI and EC mapping (ECM), a significant cantly different from baseline, with similar ratings seen at increase of brain connectivity was detected over the course 26 (72 6 8 mm) and 52 (70 6 9 mm) weeks, respectively of metreleptin treatment. A significant EC increase over (Fig. 1B). Ratings for tastiness of the test meal were be- all six measurements was found in three brain regions: tween 63 and 70 mm on the VAS and did not change hypothalamus (Montreal Neurological Institute [MNI] significantly throughout the 52 weeks of metreleptin coordinates x, y, z [in mm]: maxima 15, 23, 28, T = 4.22; treatment (data not shown). 3, 5, 211, T = 4.20; and 26, 11, 211, T = 3.24; P on Food Questionnaires cluster level = 0.040, FDR corrected), insula/superior tem- On the TFEQ, average baseline score for scale 2 (disinhi- poral gyrus (STG) (MNI coordinates local maximum 51, bition) was 7.0 6 1.0, which significantly decreased to 213, 7; T = 4.48; P on cluster level = 0.004, FDR cor- 4.2 6 1.1 as early as 4 weeks after initiation of metre- rected), and medial prefrontal cortex (mPFC) (MNI coor- leptin treatment (P , 0.05) (Fig. 2A). The average score dinates local maximum 26, 56, 16; T = 4.82; P on cluster

Figure 1—Liking ratings of food and nonfood pictures (A) and VAS for hunger and satiety (B) before, 5 min after, and 120 min after a standardized meal. *P < 0.05, **P < 0.01 compared with baseline; †P < 0.05 compared with 26 weeks as assessed by two-tailed paired Student t test. diabetes.diabetesjournals.org Schlögl and Associates 2183

Figure 2—All scales of the TFEQ (A) and 5 of 14 scales of the IEG (B) showing significant changes after initiation of metreleptin treatment. Values were obtained as described in RESEARCH DESIGN AND METHODS and Supplementary Table 1. *P < 0.05, **P < 0.01, ***P < 0.001 compared with baseline as assessed by two-tailed paired Student t test.

level , 0.001, FDR corrected) (Fig. 3). To further ascer- over 52 weeks is associated with significantly increased tain that the significant EC increases found in the midbrain resting state connectivity in the hypothalamus, insula/ cluster map the hypothalamus, a conjunction between STG, and mPFC (i.e., in both homeostatic and hedonic whole-brain ECM analysis and a hypothalamus mask brain areas). These observed effects are accompanied by created with Wake Forest University PickAtlas software significant decreases in self-reported pre- and postpran- (24) was performed. Here, voxels within the mask of dial hunger feelings as rated with a VAS (a measure for hypothalamus showed an increase of EC over time (Sup- homeostatic hunger) and food liking ratings in the fed plementary Fig. 1). The maximum of the cluster was lo- state (a measure for the hedonic perception of food). cated in 6, 2, and –17, and both peak and cluster were First, we observed a connectivity increase in the hy- significant (P , 0.05, family-wise error corrected) (Sup- pothalamus, the homeostatic center of the brain, over plementary Fig. 1). the 52 weeks of metreleptin treatment. Behavioral data Of note, a significant EC increase in all three brain showed reduced hunger ratings in the fasted state and regions was detected with both approaches of dealing increased satiety ratings after a standard meal during the with negative values in the correlation matrix (i.e., setting first 26 weeks of metreleptin treatment. These findings all negative values to zero, using the absolute value). are in accordance with independent studies on metre- When the two male subjects were excluded from the leptin treatment in patients with LD (4,13). Furthermore, analysis, the results remained similar (Supplementary Fig. TFEQ responses showed decreased scores for scale 3 (hun- 2). Furthermore, the same brain regions were also seen in ger) after initiation of metreleptin. Lower scores indicate all subjects (n = 9) when contrasting baseline (V1) against that hunger feelings (which are often perceived as dis- 52 weeks of metreleptin treatment (V6) (data not shown). turbing) are decreased (18). Of note in this context is In contrast with the linear increase over time (22.5, –1.5, that another physiological anorexigenic hormone, –0.5, 0.5, 1.5, 2.5), metreleptin treatment over 1 year also the GLP-1 agonist , had similar effects on the increased connectivity with other EC measures as calcu- hypothalamus in fMRI connectivity analysis as well as on lated with the fast ECM SPM toolbox (25), that is, in the VAS-assessed hunger and satiety in obese subjects in an- hypothalamus as assessed by norm ECM and degree ECM other study from our group (27). These results combined as well as in the insula/STG and mPFC as assessed by support the hypothesis that leptin substitution improves norm ECM and rank ECM (Supplementary Fig. 3). homeostatic satiety signaling through the hypothalamus. In connectivity analysis with region-wise pairs using However, causality cannot be established with the current the Automated Anatomical Labeling atlas (26), connectiv- design, and further investigations are required. ity for the insula/STG was also increased over the course Second, connectivity increased in the insula/STG dur- of metreleptin treatment (Supplementary Fig. 4). The hypo- ing the 52 weeks of metreleptin treatment. The insula is thalamus and the mPFC are not defined by the Automated the gustatory cortex of the brain and is involved in Anatomical Labeling atlas. By using the inverse contrast of interoception (i.e., internal sensing of food qualities like decreased EC over time, we obtained a region in the vicinity odor, taste, and nutrient composition) (28). Furthermore, of the precuneus, a part of the so-called default mode net- it is supposed to map the ongoing physiological state of work of the human brain (Supplementary Fig. 5). the body through thalamocortical pathways (29). More- over, the insula is an important region in reward processes believed to play a role in the valuation of reward, and DISCUSSION alterations are found in diseases with disrupted reward For the first time in our knowledge, this study demon- and valuation-like addictive behavior (29). In the leptin- strated that metreleptin treatment in patients with LD deficient state, the current study patients anecdotally 2184 Neural Effects of Leptin in Lipodystrophy Diabetes Volume 65, August 2016

Figure 3—Parametric contrast with increase of EC over time with six measurements (–2.5, –1.5, –0.5, 0.5, 1.5, 2.5) over 52 weeks of metreleptin treatment (yellow/orange). EC values were obtained using repeated sessions of resting state fMRI before metreleptin therapy as well as after 1, 4, 12, 26, and 52 weeks of treatment. Local maxima are given in millimeters for MNI coordinates x, y, and z. In addition, mean 6 SEM–fitted EC values at local maxima are presented for every measurement. corr, corrected. diabetes.diabetesjournals.org Schlögl and Associates 2185 described an addiction-like affinity with food, with large effects of metreleptin on brain connectivity and ana- parts of their free time revolving around food preparation lyze five time points after initiation of treatment ranging and consumption. After a meal, satiety only persisted for from 1 to 52 weeks. Furthermore, with a wide range of ;1 h. After metreleptin treatment, patients initially ex- neuropsychological assessments, we were able to differ- perienced longer periods of satiety after a meal, reduced entially picture the effects of metreleptin treatment on meal frequencies, and lost interest in thoughts about the various facets of hunger regulation and eating be- food. These descriptions are underlined by significant de- havior. However, several limitations must be discussed. creases of IEG scores for scale 1 (importance of eating), First, the study lacked a placebo or healthy control group; scale 2 (strength and triggering of desire to eat), and scale therefore, the experimental design did not allow the 11 (eating between meals). Furthermore, the behavioral distinction between metreleptin treatment effects and data indicate a stronger self-control of eating behavior nonspecific order effects. Second, measurements may after initiation of leptin substitution, as shown by IEG well have changed across repeated assessments for a scores for scale 7 (cognitive restraint on eating), which variety of reasons (e.g., habituation). Third, phase of the were significantly increased during metreleptin treat- menstrual cycle also affects eating behaviors and might ment. The score of the identically named scale 1 of the have influenced the findings. However, in only two of the TFEQ also increased but did not reach the level of statis- nine patients, intraindividual variation in sex hormone tical significance. Unfortunately, we did not assess cogni- levels can be suspected among the six measurements. tive abilities linked to the STG (30) because the behavioral Fourth, metabolic improvements in patients with LD on focus of the current study was primarily on eating regu- metreleptin therapy have already been demonstrated in lation. Overall, the findings agree with the hypothesis the past in larger samples (2); thus, the current study that metreleptin improves physiological interoception of sample size of nine patients is small for an fMRI study, food through an increase of connectivity of the insula/STG especially when taking into account that resting state compared with a leptin-deficient state. Future studies need fMRI connectivity involves numerous statistical operations to further investigate this hypothesis. on the three-dimensional volume space. However, no Third, metreleptin treatment was also associated with study with n . 1 has been published so far in metreleptin increased connectivity in the mPFC, which is a brain treatment–naive subjects with leptin-deficiency. Further- region involved in the assignment of incentive (i.e., more, the current study is the first to describe metreleptin motivational) value to food stimuli together with the treatment–associated alterations in resting state connec- orbitofrontal cortex and, thus, drives feeding behavior tivity compared with published fMRI studies with event- (31,32). In accordance with these imaging findings, we related designs ranging from n = 1 to 3 subjects (33–36); found a decrease of liking ratings of food pictures with one study reported on 10 subjects who were not treatment metreleptin therapy. TFEQ values for scale 2 (disinhibi- naive (13). A replication of the current fMRI results in tion), which indicates how vulnerable a person’s eating larger samples would be valuable. behavior is through external (e.g., smell and sight of Taken together, we have elucidated the long-term food) and internal (e.g., emotions) distractors, decreased. effects of metreleptin on brain connectivity in patients These data are supported by findings of an independent with LD. Leptin substitution causes long-term changes of study with an event-related fMRI paradigm that investi- hedonic and homeostatic central nervous networks that gated acute brain activity changes during presentation of regulate eating behavior and that are accompanied by food pictures in two adolescents with congenital leptin decreased hunger feelings and a diminished incentive deficiency (33). The authors demonstrated that positive value of food. Future studies need to assess whether correlations of brain activations in the nucleus accumbens metreleptin treatment in LD restores physiological pro- and caudate nucleus (important regions of the dopami- cesses important for the development of satiety through nergic reward system) observed in leptin deficiency were these mechanisms. not seen any more in the metreleptin-treated state, sug- gesting that leptin is necessary to suppress the incentive motivational value to food stimuli in the fed state. Funding. This work was supported by the German Research Foundation We did not find a significant correlation between (Deutsche Forschungsgemeinschaft [DFG]) CRC 1052 “Obesity Mechanisms” projects A05 to A.H., A01 to A.V. and M.S., A06 to B.P., and C06 to M.F.; the changes of the metabolic parameters HbA1c, TGs, and total, HDL, and LDL cholesterol and brain imaging results Federal Ministry of Education and Research (Bundesministerium für Bildung und (data not shown). These findings do not support the hy- Forschung [BMBF]), Germany, IFB AdiposityDiseases grant no. 01EO1501 proj- pothesis that the metabolic changes seen with metre- ects K7-83 and K7-84 to A.H. and project K6a-87 to M.F.; and the Deutsches Zentrum für Diabetesforschung (DZD) grant no. 82DZD00601. leptin significantly affect the central nervous findings. Duality of Interest. No potential conflicts of interest relevant to this article However, we cannot exclude the possibility that our meth- were reported. ods were not sensitive enough or that the sample size was Author Contributions. H.S. contributed to the study concept and design, too small. performance of experiments, data analysis, and writing and editing of the Strengths of this study include being the first fMRI manuscript. K.Mü. contributed to the study concept and design, data analysis, study in treatment-naive patients with LD to examine the and editing of the manuscript. A.H. contributed to the study concept and design, 2186 Neural Effects of Leptin in Lipodystrophy Diabetes Volume 65, August 2016

performance of experiments, data analysis, and editing of the manuscript. K.Mi. 15. Lowe MR. The importance of behavioral anchoring in neuroimaging studies contributed to the writing and editing of the manuscript. J.P. contributed to the of obesity. Am J Clin Nutr 2013;97:451–452 performance of experiments, data analysis, and editing of the manuscript. A.V. 16. Liu J, Lee J, Salazar Hernandez MA, Mazitschek R, Ozcan U. Treatment of and M.S. contributed to the study concept and design and editing of the man- obesity with celastrol. Cell 2015;161:999–1011 uscript. B.P. contributed to the study concept and design, data analysis, and 17. Stunkard AJ, Messick S. The three-factor eating questionnaire to measure writing and editing of the manuscript. M.F. contributed to the study concept and dietary restraint, disinhibition and hunger. J Psychosom Res 1985;29:71–83 design and writing and editing of the manuscript. H.S. is the guarantor of this 18. Pudel V, Westhöfer J. Fragebogen zum Ebverhalten (FEV). Göttingen, work and, as such, had full access to all the data in the study and takes Hogrefe Verlag, 1989 responsibility for the integrity of the data and the accuracy of the data analysis. 19. Diehl JM, Staufenbiel T. Inventar zum Essverhalten und Gewichtsproblemen Prior Presentation. Parts of this study were presented at the 73rd (IEG). 4th ed. Eschborn, Verlag Dietmar Klotz, 2006 Scientific Sessions of the American Diabetes Association, Chicago, IL, 21–25 20. Ashburner J, Friston KJ. Unified segmentation. Neuroimage 2005;26:839– June 2013. 851 21. Lohmann G, Müller K, Bosch V, et al. LIPSIA–a new software system for the References evaluation of functional magnetic resonance images of the human brain. Comput 1. Miehle K, Stumvoll M, Fasshauer M. Lipodystrophy. Mechanisms, clinical Med Imaging Graph 2001;25:449–457 presentation, therapy. Internist (Berl) 2011;52:362–373 [in German] 22. Brin S, Page L. Reprint of: The anatomy of a large-scale hypertextual web 2. Chong AY, Lupsa BC, Cochran EK, Gorden P. Efficacy of leptin therapy in the search engine. Comput Netw 2012;56:3825–3833 different forms of human lipodystrophy. Diabetologia 2010;53:27–35 23. Frobenius G. Über Matrizen aus nicht negativen Elementen. Sitzungsberichte 3. van Swieten MM, Pandit R, Adan RA, van der Plasse G. The neuroanatomical Preuss Akad Wiss 1912;23:456–477 function of leptin in the hypothalamus. J Chem Neuroanat 2014;61-62:207–220 24. Maldjian JA, Laurienti PJ, Kraft RA, Burdette JH. An automated method for 4. McDuffie JR, Riggs PA, Calis KA, et al. Effects of exogenous leptin on satiety neuroanatomic and cytoarchitectonic atlas-based interrogation of fMRI data sets. and satiation in patients with lipodystrophy and leptin insufficiency. J Clin Neuroimage 2003;19:1233–1239 Endocrinol Metab 2004;89:4258–4263 25. Wink AM, de Munck JC, van der Werf YD, van den Heuvel OA, Barkhof F. 5. Banks WA, Kastin AJ, Huang W, Jaspan JB, Maness LM. Leptin enters the Fast eigenvector centrality mapping of voxel-wise connectivity in functional brain by a saturable system independent of . 1996;17:305–311 magnetic resonance imaging: implementation, validation, and interpretation. 6. Satoh N, Ogawa Y, Katsuura G, et al. Satiety effect and sympathetic acti- Brain Connect 2012;2:265–274 vation of leptin are mediated by hypothalamic melanocortin system. Neurosci Lett 26. Tzourio-Mazoyer N, Landeau B, Papathanassiou D, et al. Automated ana- 1998;249:107–110 tomical labeling of activations in SPM using a macroscopic anatomical parcel- 7. Schwartz MW, Seeley RJ, Woods SC, et al. Leptin increases hypothalamic lation of the MNI MRI single-subject brain. Neuroimage 2002;15:273–289 pro-opiomelanocortin mRNA expression in the rostral arcuate nucleus. Diabetes 27. Schlögl H, Kabisch S, Horstmann A, et al. Exenatide-induced reduction in 1997;46:2119–2123 energy intake is associated with increase in hypothalamic connectivity. Diabetes 8. Wang Q, Bing C, Al-Barazanji K, et al. Interactions between leptin and Care 2013;36:1933–1940 hypothalamic neurons in the control of food intake and energy 28. Rolls ET. Taste and related systems in primates including humans. Chem homeostasis in the rat. Diabetes 1997;46:335–341 Senses 2005;30(Suppl. 1):i76–i77 9. Schloegl H, Percik R, Horstmann A, Villringer A, Stumvoll M. Peptide hor- 29. Naqvi NH, Bechara A. The insula and drug addiction: an interoceptive view mones regulating appetite–focus on neuroimaging studies in humans. Diabetes of pleasure, urges, and decision-making. Brain Struct Funct 2010;214:435–450 Metab Res Rev 2011;27:104–112 30. Bigler ED, Mortensen S, Neeley ES, et al. Superior temporal gyrus, language 10. Yu YH, Vasselli JR, Zhang Y, Mechanick JI, Korner J, Peterli R. Metabolic vs. function, and autism. Dev Neuropsychol 2007;31:217–238 hedonic obesity: a conceptual distinction and its clinical implications. Obes Rev 31. Zald DH. Orbitofrontal cortex contributions to food selection and decision 2015;16:234–247 making. Ann Behav Med 2009;38(Suppl. 1):S18–S24 11. van Zessen R, van der Plasse G, Adan RA. Contribution of the mesolimbic 32. Kringelbach ML. Food for thought: hedonic experience beyond homeostasis dopamine system in mediating the effects of leptin and ghrelin on feeding. Proc in the human brain. Neuroscience 2004;126:807–819 Nutr Soc 2012;71:435–445 33. Farooqi IS, Bullmore E, Keogh J, Gillard J, O’Rahilly S, Fletcher PC. Leptin 12. Horstmann A. The brain’s got a taste for good food. In Hedonic Eating: How regulates striatal regions and human eating behavior. Science 2007;317:1355 the Pleasurable Aspects of Food Can Affect Our Brains and Behavior. 1st ed. 34. Baicy K, London ED, Monterosso J, et al. Leptin replacement alters brain Avena NM, Ed. Oxford, Oxford University Press, 2015, p. 39–55 response to food cues in genetically leptin-deficient adults. Proc Natl Acad Sci 13. Aotani D, Ebihara K, Sawamoto N, et al. Functional magnetic resonance U S A 2007;104:18276–18279 imaging analysis of food-related brain activity in patients with lipodystrophy un- 35. Frank S, Heni M, Moss A, et al. Leptin therapy in a congenital leptin- dergoing leptin replacement therapy. J Clin Endocrinol Metab 2012;97:3663–3671 deficient patient leads to acute and long-term changes in homeostatic, reward, 14. Margulies DS, Böttger J, Long X, et al. Resting developments: a review of and food-related brain areas. J Clin Endocrinol Metab 2011;96:E1283–E1287 fMRI post-processing methodologies for spontaneous brain activity. MAGMA 36. Frank S, Heni M, Moss A, et al. Long-term stabilization effects of leptin on 2010;23:289–307 brain functions in a leptin-deficient patient. PLoS One 2013;8:e65893