lable at ScienceDirect

Clinical Nutrition xxx (2014) 1e9

Contents lists avai

Clinical Nutrition

journal homepage: ht tp: / /www.elsevier .com/locate/c lnu

Original article

Regional fat distribution in adolescent and adult females withanorexia nervosa: A longitudinal study

Marwan El Ghoch a, *, Chiara Milanese b, Simona Calugi a, Manfred J. Müller c,Maryam Pourhassan c, Antonella Ruocco a, Riccardo Dalle Grave a

a Department of Eating and Weight Disorders, Villa Garda Hospital, Via Montebaldo, 89, 37016 Garda, Verona, Italyb Department of Neurological and Movement Sciences, University of Verona, Italyc Institute of Human Nutrition and Food Science, Christian-Albrechts University, Kiel, Germany

a r t i c l e i n f o

Article history:Received 31 October 2014Accepted 14 December 2014

Keywords:DXARegional body compositionAnorexia nervosaEating disordersInpatient treatment

Abbreviations: DXA, dual-energy X-ray absorptiomioural therapy; BMI, body mass index; kg, kilogram;* Corresponding author. Department of Eating and

Hospital, Via Montebaldo, 89, I-37016 Garda, VR, Itfax: þ39 045 8102884.

E-mail address: marwan1979@hotmail.com (M. El

http://dx.doi.org/10.1016/j.clnu.2014.12.0120261-5614/© 2014 Elsevier Ltd and European Society

Please cite this article in press as: El Ghoclongitudinal study, Clinical Nutrition (2014)

s u m m a r y

Background & aims: No study has yet compared body fat distribution before and after weight restorationin adolescent and adult patients with anorexia nervosa (AN) treated according to the same protocol. Thestudy was set up to measure body fat distribution before and after short-term weight restoration inadolescent and adult patients with AN treated according to the same protocol in a specialist inpatientunit.Methods: We recruited 33 consecutive adolescent female patients with AN, and 33 controls matched byage and post-treatment BMI centile, as well as 33 adult female patients with AN, and 33 controlsmatched by age and post-treatment BMI. Dual-energy X-ray absorptiometry (DXA) was used to assessbody composition before and after short-term weight restoration (BMI � 18.5 kg/m2).Results: Compared with controls, both adolescents and adults with AN showed that a greater amount offat was lost from the extremities than the trunk before weight restoration, and that there was a centraladiposity phenotype after short-term weight restoration. There were no significant differences in bodyfat distribution between adolescents and adults with AN before or after short-term weight restoration.Conclusions: Adolescent and adult females with AN have similar body fat distribution both before andafter short-term weight restoration, and show a central adiposity phenotype after short-term weightrestoration. The clinical implications of this finding are as yet unknown.Clinical Trials Registryd Regional Fat Distribution in Adolescent Girls and Adults with Anorexia Nervosa(ISRCTN73572502).

© 2014 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism. All rights reserved.

1. Introduction

The major clinical features of the eating disorder anorexianervosa (AN) are loss of weight and reduction of lean and fatmasses [1e3]. Although weight restoration is considered the keystrategy of treatment [4], recent studies suggest that change inbody composition may also influence the outcome of this disorder.For example, two studies showed an association between poorlong-term outcome and a lower total fat percentage after short-

etry; CBT, cognitive behav-kcal, kilocalorie.Weight Disorders, Villa Gardaaly. Tel.: þ39 045 8103915;

Ghoch).

for Clinical Nutrition and Metabol

h M, et al., Regional fat dist, http://dx.doi.org/10.1016/j.c

term weight restoration in females with AN [5,6], while otherstudies found that regional fat distribution abnormalities arerelated to lower physical fitness [7] and negative metabolic con-sequences [8]. If confirmed, these findings indicate that studyingbody composition before and after weight restoration in patientswith AN, aside from being of interest to physiologists, could havesome important clinical implications.

Both body composition and body fat distribution before andafter weight restoration have beenwidely assessed in adult femaleswith AN, and four main conclusions can be drawn, namely [9]: (i)before weight restoration patients have lower peripheral thancentral (trunk) body fat; (ii) there is an association between short-term weight restoration and a central adiposity phenotype[8,10,11]; (iii) there is an association between central fat distribu-tion and increased insulin resistance [8], but this causes no changein psychological distress levels, nor does it have an adverse effect on

ism. All rights reserved.

ribution in adolescent and adult females with anorexia nervosa: Alnu.2014.12.012

M. El Ghoch et al. / Clinical Nutrition xxx (2014) 1e92

the psychopathology of the eating disorder [11]; and (iv) thereappears to be a normalization in the centralization of body fatduring long-term maintenance of weight [12].

Fewer studies have been conducted in adolescents with AN, butavailable data [13] indicate that they tend to lose more central thanperipheral fat beforeweight restoration [14e16], anddonot developa central adiposity phenotype during weight restoration [13].However, due to a paucity of data, no definitive conclusions can yetbe drawn about the differences between adolescents and adultswith AN in terms of body composition before and after weightrestoration. Indeed, the research conducted on adolescents with ANtodatehas suffered fromvariousmethodological limitations, suchascross sectional designs [16], small sample groups and subgroups[13,14], incomplete weight restoration [17], and lack of a controlgroup [17]. Furthermore, studies have differed widely in terms ofdesign andmeasurement timing [9]. Last, but by no means least, nostudies have yet compared adolescent and adult populations withAN assessed and treated according to an identical protocol.

Hence, the aims of our study were: (i) to assess regional fatdistribution in adolescent and adult females with AN and controlsmatched by age and post-treatment Body Mass Index (BMI) or BMIcentile; (ii) to assess changes in these parameters after short-termweight restoration (i.e., BMI � 18.5 kg/m2); and (iii) to comparebody composition of adolescent and adult females with AN treatedusing the same protocol and a uniform methodology (i.e., bodycomposition assessment).

2. Methods

2.1. Participants

We recruited 33 consecutive adolescent females with AN, 33adolescent controls matched by age and post-treatment BMI cen-tiles, 33 adult females with AN, and 33 adult controls matched byage and post-treatment BMI. Patients were consecutive andvoluntarily admitted to the Villa Garda Hospital eating disorderinpatient unit from 2012 to 2014. The following inclusion criteriawere applied: (i) age, 13e19 years for adolescents [18] and �20years for adults; (ii) baseline BMI � 18.5 kg/m2 in adults, or thecorresponding BMI centile in adolescents; (iii) diagnosis of AN ac-cording to DSM-5 criteria [19]; (iv) patients whose eating disorderwas deemed too severe to be managed in an outpatient pro-gramme, and those who had previously been unsuccessfullytreated via a less intensive outpatient approach; and (v) post-treatment BMI � 18.5 kg/m2 in adults, or the corresponding BMIcentile in adolescents. Patients with any of the following wereexcluded: schizophrenia, any other psychotic disorders, or currentsubstance misuse. School and university listservs and advertise-ments were used to randomly recruit control subjects, whowere allin good health and of stable weight, reporting regular menstrua-tion. Exclusion criteria for the control group comprised: history ofeating disorders, other significant psychiatric or medical condi-tions, and ongoing psychotropic treatment. Eligibility (inclusionand exclusion criteria) of both patients and controls was deter-mined by means of interviews conducted by an expert in the field(RDG).

The Institutional Review Board of Villa Garda Hospital, Verona,reviewed and approved the study design. All participants (or theirlegal guardians if under 18) provided informed written consent fortheir personal information to be used anonymously.

2.2. Inpatient treatment protocol

More details regarding the inpatient treatment can be foundelsewhere [20], but, in brief, it comprises an adapted version of

Please cite this article in press as: El Ghoch M, et al., Regional fat distlongitudinal study, Clinical Nutrition (2014), http://dx.doi.org/10.1016/j.c

CBT-E e an enhanced form of cognitive behavioural therapydesigned for eating disorders [21]. Patients are treated for 13 weeksas inpatients, then 7 weeks in day hospital, making a total of 20weeks. Patients attend both individual and group sessions in whichall the main strategies and procedures of CBT-E are delivered. Pa-tients also receive dietician-assisted eating in the first weeks of theprogramme, a strategy not foreseen in the original outpatient CBT-E. Assisted eating is continued until patients' BMI reaches�18.5 kg/m2 in adults (the corresponding centile in adolescents). This isachieved by increasing the dietary energy content given to patientsfrom 1500 to 2500 kcals in 500 kcal/week increments. Subse-quently the dietary intake is adjusted so that a steady weight gainrate of 1e1.5 kg/week is achieved. When BMI reaches � 19.0 inadults (the corresponding centile in adolescents), the daily caloriecontent is continually adjusted to maintain a body weight within a2-kg range of this target. The diet was designed to conform to theItalian National Guidelines for Healthy Eating [22] and contain allthe main food groups. Adult patients of BMI� 15.0 and adolescentswith the corresponding BMI centile of stable medical condition alsoattended 60-min physiotherapist-led exercise sessions twice perweek. Exercises included calisthenics for posture and to helprestore muscle strength and flexibility, and aerobics for cardiovas-cular fitness. No patients were prescribed any psychotropic medi-cation, and regimens (i.e., selective serotonin reuptake inhibitors)that were ongoing at admission (by 12 patients in both groups)were gradually phased out in the first 2 weeks of the programme.

2.3. Measurements

Collection of data was performed during the first week ofinpatient treatment, and in the final week of day hospital.

2.3.1. Body weight and heightA medical doctor affiliated to the study measured body weight

and height by means of medical weighing scales and a stadiometer,respectively. Weighing was conducted before breakfast with par-ticipants without shoes and wearing only underwear. Adult BMIwas calculated using the standard formula, i.e., body weight in ki-lograms divided by height in metres squared. Adolescent BMIcentiles were calculated according to the growth charts provided bythe Centre for Disease Control and Prevention (www.cdc.gov/growthcharts).

2.3.2. Body compositionA dual-energy x-ray absorptiometry (DXA) scanner and soft-

ware (Prodigy Primo Lunar, A223040501, General Electric Com-pany, Madison. WI 53707-7550, USA-EnCORE TM 2009 (v13.31)software) was used to measure total and regional fat, and lean andbone masses, as described previously [11]. In brief, scans wereperformed in the morning with no special preparation. Participantswere scanned in their underwear and without any metallic jewel-lery, etc. Fat masses were measured in the arm, trunk, leg, androidand gynoid regions. The whole body DXA was used to calculatepercentage trunk and extremity fat, and the trunk-fat-to-extremity-fat ratio, via the following formulas:

1. Total fat mass (FM): total FM in kg.2. Total FM percentage (%): total FM/total body weight � 100.3. Total lean mass: total lean mass in kg.4. Total lean mass percentage (%): total lean mass/total body

weight � 100.5. Trunk fat percentage ¼ trunk fat/total body fat � 100.6. Extremity fat percentage ¼ (arms fat þ legs fat)/total body

fat � 100.

ribution in adolescent and adult females with anorexia nervosa: Alnu.2014.12.012

Table 2Body weight and body composition variables in pre- and post-treatmenta adolescent patients with anorexia nervosa (AN) and control group.

Comparison betweenPre-treatment AN andcontrol group

Comparison betweenpost-treatment AN andcontrol group

Comparison betweenpre- and post-treatment AN values

Pre-treatment AN(N ¼ 33)

Post- treatment AN(N ¼ 33)

Control group(N ¼ 33)

Mean difference Pb Mean difference Pb Mean change Pc

Weight (kg) 38.9 ± 5.1 49.4 ± 3.9 51.7 ± 5.5 �12.75 <0.001 �2.24 0.062 10.51 <0.001BMI centile 1.6 ± 4.0 31.7 ± 10.6 32.2 ± 11.4 �30.55 <0.001 �0.53 0.847 30.03 <0.001Total body compositionTotal FM (kg) 4.1 ± 2.8 12.2 ± 2.7 12.1 ± 2.9 �7.97 <0.001 0.17 0.800 8.14 <0.001Total FM (%) 10.1 ± 5.9 24.8 ± 5.3 23.1 ± 3.8 �13.06 <0.001 1.65 0.155 14.71 <0.001Total lean mass (kg) 32.8 ± 3.5 35.1 ± 3.8 37.2 ± 3.4 �4.37 <0.001 �2.04 0.025 2.33 <0.001Total lean mass (%) 84.7 ± 5.6 71.1 ± 5.0 72.1 ± 3.4 12.56 <0.001 �1.08 0.316 �13.64 <0.001Lean-mass-to-FM ratio 11.8 ± 6.9 3.1 ± 1.0 3.2 ± 0.7 8.60 <0.001 �0.16 0.453 �8.76 <0.001Fat mass distributionExtremity fat percentage (%) 50.0 ± 6.7 48.9 ± 5.4 57.4 ± 3.3 �7.37 <0.001 �8.49 <0.001 �1.11 0.212Trunk fat percentage (%) 40.5 ± 5.5 46.7 ± 5.4 35.6 ± 3.4 4.88 <0.001 11.07 <0.001 6.19 <0.001Trunk-fat-to-extremity-fat ratio 0.8 ± 0.2 1.0 ± 0.2 0.6 ± 0.1 0.21 <0.001 0.35 <0.001 0.14 <0.001

All values are mean ± SD.a Pre-treatment assessment was conducted in the first week after inpatient admission, and post-treatment assessment in the last week of the 20-week treatment pro-

gramme, which comprised 13 weeks of inpatient therapy followed by 7 weeks of partial hospitalization. FM: fat mass Holm's procedure was applied to adjust P values formultiple comparisons.

b P values were determined using the t-test for independent samples.c P values were determined using the t-test for paired samples.

Table 3Body weight and body composition variables in pre- and post-treatmenta adult patients with anorexia nervosa (AN) and control group.

Comparison betweenPre-treatment AN andcontrol group

Comparison betweenpost-treatment AN andcontrol group

Comparison betweenPre- and post-treatment AN values

Pre-treatment AN(N ¼ 33)

Post- treatment AN(N ¼ 33)

Control group(N ¼ 33)

Mean difference Pb Mean difference Pb Mean change Pc

Weight (kg) 38.7 ± 5.3 50.1 ± 3.6 51.3 ± 3.1 �12.58 <0.001 �1.23 0.143 11.35 <0.001BMI (kg/m2) 15.1 ± 1.8 19.6 ± 0.7 19.6 ± 0.7 �4.46 <0.001 �0.03 0.868 4.43 <0.001Total body compositionTotal FM 4.0 ± 3.0 12.1 ± 3.3 11.8 ± 1.8 �7.77 <0.001 0.31 0.636 8.09 <0.001Total FM (%) 9.8 ± 6.6 24.1 ± 6.2 23.0 ± 3.6 �13.14 <0.001 1.11 0.379 14.25 <0.001Total lean mass 32.9 ± 4.0 36.1 ± 4.0 37.5 ± 3.2 �4.57 <0.001 �1.39 0.122 3.18 <0.001Total lean mass (%) 85.4 ± 6.2 72.1 ± 6.3 73.0 ± 3.5 12.37 <0.001 �0.92 0.466 �13.29 <0.001Lean-mass-to-FM ratio 13.6 ± 8.2 3.4 ± 1.7 3.3 ± 0.8 10.28 <0.001 0.07 0.832 �10.21 <0.001Fat mass distributionExtremity fat percentage (%) 48.8 ± 10.2 47.8 ± 7.2 57.8 ± 4.5 �8.98 <0.001 �9.94 <0.001 �0.96 0.363Trunk fat percentage (%) 41.8 ± 8.5 47.7 ± 7.2 35.4 ± 4.5 6.44 <0.001 12.27 <0.001 5.83 <0.001Trunk fat-to-extremity-fat ratio 0.9 ± 0.4 1.0 ± 0.3 0.6 ± 0.1 0.31 <0.001 0.42 <0.001 0.11 0.006

All values are mean ± SD.a Pre-treatment assessment was conducted in the first week after inpatient admission, and post-treatment assessment in the last week of the 20-week treatment pro-

gramme, which comprised 13 weeks of inpatient therapy followed by 7 weeks of partial hospitalization. FM: fat mass. Holm's procedure was applied to adjust P values formultiple comparisons.

b P values were determined using the t-test for independent samples.c P values were determined using the t-test for paired samples.

Table 1Demographic and baseline clinical characteristics of patients with anorexia nervosa (AN) and control groups.

Pre-treatment adolescent with AN(N ¼ 33)

Adolescent control group (N¼ 33) Pre-treatment adults with AN (N¼ 33) Adult controlgroup(N ¼ 33)

Age (y) 16.0 ± 1.5 16.3 ± 1.8 30.6 ± 6.7 30.6 ± 6.9Highest educational qualification, n (%)Elementary school 2 (6.1) 4 (12.1) 0 0Middle school certificate 31 (93.9) 28 (84.8) 3 (9.1) 2 (6.1)High school diploma 0 1 (3.0) 24 (72.7) 23 (69.7)Bachelor's degree 0 0 6 (18.2) 6 (18.2)Missing data 0 0 e 2 (6.1)

Presence of amenorrhoea, n(%)

31 (93.3) e 26 (78.8) e

Duration of illness (years)a 1.8 ± 1.4 e 11.7 ± 7.2 e

Amount of exercise (minutes)b 1385.4 ± 2109.4 e 1565.0 ± 3443.7 e

All values are presented as mean ± SD or number (%).a Data not available for the control group.b In the last 28 days (assessed using the 12th edition of the Eating Disorder Examination interview).

M. El Ghoch et al. / Clinical Nutrition xxx (2014) 1e9 3

Please cite this article in press as: El Ghoch M, et al., Regional fat distribution in adolescent and adult females with anorexia nervosa: Alongitudinal study, Clinical Nutrition (2014), http://dx.doi.org/10.1016/j.clnu.2014.12.012

Table 4Comparison of adolescent and adult anorexia nervosa (AN) patients' pre- and post-treatmenta fat mass distribution variables.

Comparison betweenpre-treatment ANgroups

Comparison betweenPost-treatment ANgroups

Pre-treatmentadolescent AN(N ¼ 33)

Post- treatment adolescent AN(N ¼ 33)

Pre-treatmentadult AN(N ¼ 33)

Post-treatmentadult AN(N ¼ 33)

Mean difference Pb Mean difference Pb

Fat mass distributionExtremity fat percentage (%) 50.0 ± 6.7 48.9 ± 5.4 48.8 ± 10.2 47.8 ± 7.2 1.23 0.564 1.07 0.493Trunk fat percentage (%) 40.5 ± 5.5 46.7 ± 5.4 41.8 ± 8.5 47.7 ± 7.2 �1.33 0.454 �0.97 0.539Trunk-fat-to-extremity-fat ratio 0.8 ± 0.2 1.0 ± 0.2 0.9 ± 0.4 1.0 ± 0.3 �0.10 0.210 �0.07 0.346

All values are mean ± SD.P values were determined using the t-test for paired samples.

a Pre-treatment assessment was conducted in the first week after inpatient admission, and post-treatment assessment in the last week of the 20-week treatment pro-gramme, which comprised 13 weeks of inpatient therapy followed by 7 weeks of partial hospitalization. Holm's procedure was applied to adjust P values for multiplecomparisons.

b P values were determined using the t-test for independent samples.

Fig. 1. Relationships between pre-treatment (T0) body weight vs. a) fat mass (FM) and b) lean body mass, and post-treatment (T1) body weight vs. c) fat mass (FM) and d) lean bodymass in anorexia nervosa group (AN, adults plus adolescents, n ¼ 66; open symbols) and control group (adults plus adolescents, n ¼ 66; closed symbols). Solid lines are regressionlines.

M. El Ghoch et al. / Clinical Nutrition xxx (2014) 1e94

Please cite this article in press as: El Ghoch M, et al., Regional fat distribution in adolescent and adult females with anorexia nervosa: Alongitudinal study, Clinical Nutrition (2014), http://dx.doi.org/10.1016/j.clnu.2014.12.012

Fig. 2. Relationships between pre-treatment (T0) fat mass (FM) vs. trunk fat and extremity fat in a) adults and b) adolescents with anorexia nervosa (AN, n ¼ 33; open symbols) andcontrol groups (n ¼ 33; closed symbols). Solid lines are regression lines.

M. El Ghoch et al. / Clinical Nutrition xxx (2014) 1e9 5

7. Trunk-fat-to-extremity-fat ratio ¼ Trunk fat percentage/Ex-tremity fat percentage.

8. Lean-mass-to-FM ratio: Total lean mass/total FM.

2.4. Statistical analysis

Reported data are all expressed as mean ± SD or number (per-centage). Data was analysed using SPSS Statistics software (SPSS,version 20.0; IBM). The t-test for independent samples was used tocompare demographic and clinical features of adolescent and adultpatients with AN, and their respective controls. The significance ofdifferences in clinical variables from admission to discharge inadolescent and adult patients with ANwere analysed bymeans of apaired-samples t-test. Relationships between clinical variableswere assessed by means of Pearson's correlation coefficient, andsignificantly correlated variables were further analysed via singlelinear regression analysis.

Please cite this article in press as: El Ghoch M, et al., Regional fat distlongitudinal study, Clinical Nutrition (2014), http://dx.doi.org/10.1016/j.c

To adjust P values for multiple comparisons, we applied Holm'sprocedure, which is a less conservative and more powerful methodthan Bonferroni's test [23,24]. All tests were 2-tailed, and P < 0.025was accepted as the significance threshold.

3. Results

Table 1 shows the demographic and baseline clinical charac-teristics of patients with AN and controls. No differences werefound between adolescent patients and controls or between adultpatients and controls. However, as expected, adolescent and adultpatients with AN had significant differences in terms of educationallevel, age and duration of illness (all P < 0.001). Nevertheless, theyreported similar percentage presence of amenorrhoea, and similaramounts of exercise measured in the 28 days preceding admission.

Tables 2 and 3, respectively, show the body composition vari-ables of adolescent patients with AN and controls, and adult

ribution in adolescent and adult females with anorexia nervosa: Alnu.2014.12.012

Fig. 3. Relationships between post-treatment (T1) fat mass (FM) vs. trunk fat and extremity fat in a) adults and b) adolescents with anorexia nervosa (AN, n ¼ 33; open symbols)and control groups (n ¼ 33; closed symbols). Solid lines are regression lines.

M. El Ghoch et al. / Clinical Nutrition xxx (2014) 1e96

patients with AN and controls. At baseline, adolescent and adultpatients with AN had significantly lower weight, BMI centile orBMI, total FM, total lean mass and extremity fat percentages, and ahigher total lean mass percentage, lean-mass-to-FM ratio, trunk fatpercentage and trunk-fat-to-extremity-fat ratio than their respec-tive controls. No significant differences between adolescent andadult patients with AN were detected in any of the body fat dis-tribution variables assessed (Table 4). Moreover at baseline, for agiven body weight, both adolescents and adults with AN showedlower total FM, total lean mass (Fig. 1a, b), and lower trunk fat andextremity fat than their control groups. (Figure 2a for adults; Fig. 2bfor adolescents).

A significant increase in the body weight of both adolescent andadult patients was seen from baseline to the end of treatment. Thiswas equivalent to a BMI centile increase of 30 (P < 0.001) inadolescent patients and a BMI increase of 4.4 (P < 0.001) in adultpatients. In adolescents, the mean weight gain was 10.5 ± 3.3 kg,comprising 8.2 ± 2.7 kg FM and 2.3 ± 1.6 kg total lean mass(Table 2). In adults, the mean weight gain was 11.3 ± 4.8 kg,comprising 8.1 ± 2.8 kg FM and 3.2 ± 3.7 kg total lean mass

Please cite this article in press as: El Ghoch M, et al., Regional fat distlongitudinal study, Clinical Nutrition (2014), http://dx.doi.org/10.1016/j.c

(Table 3). Taking the AN group as a whole, for a given body weight,total FM and total lean mass were similar to control values (Fig. 1c,d).

Comparing pre-treatment and post-treatment body composi-tion variables revealed that all underwent significant improvementin both adolescent and adult patients with AN over the course oftreatment, except for extremity fat percentage, which did notchange significantly (Tables 2 and 3).

There were no significant differences between adolescent con-trols and post-treatment adolescent patients with AN detected ineither weight, BMI centile, total and percentage fat and leanmasses, or lean-mass-to-FM ratio (Table 2). Similar results wereyielded by a post-treatment comparison of adult patients with ANand controls, with no significant differences revealed in eitherweight, BMI, total and percentage fat and lean masses, or lean-mass-to-FM ratio (Table 3).

Nonetheless, comparedwith controls, both adolescent and adultpatients with AN had lower extremity-fat percentage, higher trunk-fat percentage and higher trunk-fat-to-extremity-fat ratio (Tables 2and 3) after weight restoration, displaying a central adiposity

ribution in adolescent and adult females with anorexia nervosa: Alnu.2014.12.012

Fig. 4. Relationships between body weight gain vs. a) fat mass gain (FM) and b) lean body mass gain, and FM gain vs. c) gain in trunk fat and d) gain in extremity fat in adults(n ¼ 33; closed symbols) and adolescents (n ¼ 33; open symbols) with anorexia nervosa (AN). Solid lines are regression lines.

M. El Ghoch et al. / Clinical Nutrition xxx (2014) 1e9 7

phenotype (Fig. 3a for adults; Fig. 3b for adolescents). No significantdifferences between adolescent and adult patients were foundpost-treatment (Table 4).

Hence, upon weight gain, significant increases in FM and leanmass were observed in both adolescents and adults with AN.However adolescents with AN tended to gainmore FM and less leanmass with respect to adults with AN (Fig. 4a, b). In addition, theproportion of FM in the weight gain was found to be directlycorrelated to the increase in either trunk fat or extremity fat (Fig. 4c,d).

4. Discussion

Assessment of body composition and fat distribution via DXA in66 patients with AN, divided into two groups e adolescents andadults e before and after short-term weight restoration, and 66healthy controls matched by age and patients' post-treatment BMIor BMI centile yielded three major findings. The first was thatbefore weight restoration, both adolescents and adults with ANdisplayed lower body weight, BMI centile or BMI, total fat

Please cite this article in press as: El Ghoch M, et al., Regional fat distlongitudinal study, Clinical Nutrition (2014), http://dx.doi.org/10.1016/j.c

percentage, total FM, and total lean mass than controls, but thatthese variables did not significantly differ between the patients andtheir respective controls after weight restoration. This indicatesthat patients with AN completely restored their fat mass during the20 weeks of treatment.

The second finding was that, although adolescents tend to gainmore total FM compared to adults, a phenomenon that might beattributed to the pubertal growth spurt [25], no significant differ-ences were found in the measured body fat distribution variablesbetween the two group of patients either before or after weightrestoration. This is of particular interest, as no study had previouslycompared body composition in adolescent and adult patients withAN assessed and treated according to an identical protocol beforeand after weight restoration.

The third finding was that before weight restoration, both adultand adolescent patients with AN appear to have greater fat loss inthe extremities than the trunk, while after short-term weightrestoration they display a central adiposity phenotype with respectto healthy age-matched controls. Our data for adult patients withAN confirm previous observations [8,10,11]. Our adolescent data

ribution in adolescent and adult females with anorexia nervosa: Alnu.2014.12.012

M. El Ghoch et al. / Clinical Nutrition xxx (2014) 1e98

contrast with previous reports which did not find a greater loss oftrunk than extremity fat before weight restoration [14e16], or anyincrease in central adiposity after weight restoration [13,14]. Theseconflicting results may be explained by the differences betweenstudies in the timing of body composition assessment (short-termversus long-term) [13,14], as well as differences in the age limitinclusion range, which varied from 12 to 22 years in previousstudies [13e17]. Finally, in some previous studies, control groupswere matched by age but not by patients' post-treatment BMIcentile [14], whichmay have a great influence on the interpretationof results.

To our knowledge, our study is the first to report the occurrenceof the central adiposity phenotype in adolescent, as well as adult,females with AN after short-termweight restoration. If this findingis confirmed, the hypothesis that trunk fat accumulation may havea significant negative metabolic effect during short-term weightrestoration should be investigated by future studies in adolescentpatients, as performed recently in adult females with AN [8,11].

The first main strength of this study is that, to our knowledge, itis the first to investigate how short-term weight restoration affectsbody fat distribution in adolescent and adult patients with AN andtheir respective control groups matched by age and post-treatmentBMI centile or BMI. Second, body composition (i.e., total bodycomposition, regional body fat percentages and FM distribution)was measured using DXA software, which although not yetaccepted as the gold standard, is known to exhibit a high level ofprecision in measuring the body composition, particularly body fat,of patients with AN [26]. Third, our study is the first to assessadolescent and adult females with AN given identical treatments(i.e., same setting and duration of treatment, similar weight regainrate, and nutritional and physical activity programmes) using auniform methodology (body composition measurement).

Our study does, however, have several limitations. The majorlimitation is that only one inpatient unit was the setting for datacollection from patients treated via a single treatment programme,so external validation is required. It is also possible that they ourfindings may not apply to patients with AN who regain weightmore slowly, plausibly resulting in a different regional fat accu-mulation pattern. Second, no biochemical or biological testing wasconducted, either before or after weight restoration, to clarify theimplications of the observed changes in body fat distribution inadolescent and adult patients with AN. Third, we did not assessphysical activity levels in the control groups, which are known tohave an impact on weight and body composition. Fourth, wecalculated body composition changes after short-term weightrestoration but, as reported by some studies in both adolescents[13,27] and adults [12], it is likely that the preferential distributionof body fat in the central region is a transitory phenomenon, andbody fat distribution may normalize during long-term mainte-nance of restored weight. Nevertheless, these data require confir-mation. Finally, lean mass may be underestimated by DXA [28],particularly in underweight patients with AN. Indeed, Haas et al.showed that patients whose lean mass reduction is fairly modestmay suffer from severe protein depletion [29]. This is due to anassociation between starvation-induced changes in body compo-sition and an increase in total body water, which may mask lowtotal body protein in DXA's 3-compartment model. Direct methodsmay therefore be more reliable in terms of lean mass measurementin underweight patients.

As a whole, the data we obtained show that adolescents andadults with AN share similar body fat distribution patterns bothbefore and after they regain weight. With respect to healthy age-matched controls, both groups of patients appear to show greaterfat loss in the extremities than the trunk before weight restoration,and a central adiposity phenotype after short-term weight

Please cite this article in press as: El Ghoch M, et al., Regional fat distlongitudinal study, Clinical Nutrition (2014), http://dx.doi.org/10.1016/j.c

restoration. Future studies should investigate, both in adolescentand adult patients with AN, if the abnormal body fat distributionafter short-term weight restoration could have some potentialnegative metabolic and/or psychological effects.

Statement of authorship

All co-authors claim authorship in that they have made sub-stantial contributions to the drafting and approval of this paper,and/or to the design and execution of the study itself.

Source of funding

The authors are aware of no affiliations or financial involvement,and receive no funding from any organization that has a vestedinterest in or a conflict with the topic or equipment discussedherein.

Conflict of interest

All authors declare no competing financial interests in relationto the work described.

Acknowledgements

The study was designed by ME and RD. Data assessment wascarried out byME, CM and AR. Statistical analysis was performed bySC and MP. ME, RD, SC, MP and MM co-wrote the paper. MM andMP were responsible for the final revision. All authors read andapproved the final manuscript.

References

[1] El Ghoch M, Alberti M, Milanese C, Battistini NC, Pellegrini M, Capelli C, et al.Comparison between dual-energy X-ray absorptiometry and skinfolds thick-ness in assessing body fat in anorexia nervosa before and after weightrestoration. Clin Nutr 2012;31:911e6.

[2] Haas V, Stark D, Kohn M, Muller MJ, Clarke S, Blumenthal C, et al. Evaluation ofmethods to assess reduced body fat in patients with anorexia nervosa. J AmColl Nutr 2013;32:2e10.

[3] Haas V, Onur S, Paul T, Nutzinger DO, Bosy-Westphal A, Hauer M, et al. Leptinand body weight regulation in patients with anorexia nervosa before andduring weight recovery. Am J Clin Nutr 2005;81:889e96.

[4] Marzola E, Nasser JA, Hashim SA, Shih PA, Kaye WH. Nutritional rehabilitationin anorexia nervosa: review of the literature and implications for treatment.BMC Psychiatry 2013;13:290.

[5] Mayer LE, Roberto CA, Glasofer DR, Etu SF, Gallagher D, Wang J, et al. Doespercent body fat predict outcome in anorexia nervosa? Am J Psychiatry2007;164:970e2.

[6] Bodell LP, Mayer LE. Percent body fat is a risk factor for relapse in anorexianervosa: a replication study. Int J Eat Disord 2011;44:118e23.

[7] Alberti M, Galvani C, Capelli C, Lanza M, El Ghoch M, Calugi S, et al. Physicalfitness before and after weight restoration in anorexia nervosa. J Sports MedPhys Fit 2013;53:396e402.

[8] Prioletta A, Muscogiuri G, Sorice GP, Lassandro AP, Mezza T, Policola C, et al. Inanorexia nervosa, even a small increase in abdominal fat is responsible for theappearance of insulin resistance. Clin Endocrinol (Oxf) 2011;75:202e6.

[9] El Ghoch M, Calugi S, Lamburghini S, Dalle Grave R. Anorexia nervosa andbody fat distribution: a systematic review. Nutrients 2014;6:3895e912.

[10] Mayer L, Walsh BT, Pierson Jr RN, Heymsfield SB, Gallagher D, Wang J, et al.Body fat redistribution after weight gain in women with anorexia nervosa. AmJ Clin Nutr 2005;81:1286e91.

[11] El Ghoch M, Milanese C, Calugi S, Pellegrini M, Battistini NC, Dalle Grave R.Body composition, eating disorder psychopathology, and psychologicaldistress in anorexia nervosa: a longitudinal study. Am J Clin Nutr 2014;99:771e8.

[12] Mayer LE, Klein DA, Black E, Attia E, Shen W, Mao X, et al. Adipose tissuedistribution after weight restoration and weight maintenance in women withanorexia nervosa. Am J Clin Nutr 2009;90:1132e7.

[13] de Alvaro MT, Munoz-Calvo MT, Barrios V, Martinez G, Martos-Moreno GA,Hawkins F, et al. Regional fat distribution in adolescents with anorexianervosa: effect of duration of malnutrition and weight recovery. Eur J Endo-crinol 2007;157:473e9.

ribution in adolescent and adult females with anorexia nervosa: Alnu.2014.12.012

M. El Ghoch et al. / Clinical Nutrition xxx (2014) 1e9 9

[14] Misra M, Soyka LA, Miller KK, Grinspoon S, Levitsky LL, Klibanski A. Regionalbody composition in adolescents with anorexia nervosa and changes withweight recovery. Am J Clin Nutr 2003;77:1361e7.

[15] Misra M, Miller KK, Almazan C, Worley M, Herzog DB, Klibanski A. Hormonaldeterminants of regional body composition in adolescent girls with anorexianervosa and controls. J Clin Endocrinol Metab 2005;90:2580e7.

[16] Kerruish KP, O'Connor J, Humphries IR, Kohn MR, Clarke SD, Briody JN, et al.Body composition in adolescents with anorexia nervosa. Am J Clin Nutr2002;75:31e7.

[17] Franzoni E, Ciccarese F, Di Pietro E, Facchini G, Moscano F, Iero L, et al. Follow-up of bone mineral density and body composition in adolescents withrestrictive anorexia nervosa: role of dual-energy X-ray absorptiometry. Eur JClin Nutr 2014;68:247e52.

[18] Age limits and adolescents paediatr child health8; 2003. p. 577e8.[19] American Psychiatric Association. Diagnostic and statistical manual of

mental disorders. 5thed. Arlington: American Psychiatric Publishing; 2013.DSM-5™.

[20] Dalle Grave R. Multistep cognitive behavioral therapy for eating disorders:theory, practice, and clinical cases. New York, NY: Jason Aronson; 2013.

[21] Fairburn C. Cognitive behavior therapy and eating disorders.. New York, NY:Guilford Press; 2008.

Please cite this article in press as: El Ghoch M, et al., Regional fat distlongitudinal study, Clinical Nutrition (2014), http://dx.doi.org/10.1016/j.c

[22] Istituto Nazionale di Ricerca per gli Alimenti e la Nutrizione (INRAN). Italianguidelines for healthy eating [Linee guida per una sana alimentazione ital-iana]. In Italian. 2003.

[23] Aickin M, Gensler H. Adjusting for multiple testing when reporting researchresults: the Bonferroni vs Holm methods. Am J Public Health 1996;86:726e8.

[24] Holm S. A simple sequentially rejective multiple test procedure. Scand J Stat1979;6:6.

[25] Vizmanos B, Marti-Henneberg C. Puberty begins with a characteristic subcu-taneous body fat mass in each sex. Eur J Clin Nutr 2000;54:203e8.

[26] Bredella MA, Ghomi RH, Thomas BJ, Torriani M, Brick DJ, Gerweck AV, et al.Comparison of DXA and CT in the assessment of body composition in pre-menopausal women with obesity and anorexia nervosa. Obes (Silver Spring)2010;18:2227e33.

[27] Misra M, Shulman D, Weiss A. Fact sheet. Anorex J Clin Endocrinol Metab2013;98:35Ae6A.

[28] Haas VK, Allen JR, Kohn MR, Clarke SD, Zhang S, Briody JN, et al. Total bodyprotein inhealthy adolescent girls: validationof estimates derived fromsimplermeasures with neutron activation analysis. Am J Clin Nutr 2007;85:66e72.

[29] Haas VK, Kohn MR, Clarke SD, Allen JR, Madden S, Muller MJ, et al. Bodycomposition changes in female adolescents with anorexia nervosa. Am J ClinNutr 2009;89:1005e10.

ribution in adolescent and adult females with anorexia nervosa: Alnu.2014.12.012