Anthropometry, CT, and DXA as predictors of GH deficiency in premenopausal women: ROC curve analysis

Anthropometry, CT, and DXA as predictors of GH deficiency in premenopausal

women: ROC curve analysis

Miriam A. Bredella1, Andrea L. Utz2, Martin Torriani1, Bijoy Thomas1, David A.

Schoenfeld3, Karen K. Miller2

1Department of Radiology, Massachusetts General Hospital, Yawkey 6 E, 55 Fruit Street

Boston, MA 02114

2Neuroendocrine Unit, Massachusetts General Hospital, Bulfinch 457B, 55 Fruit Street

Boston, MA 02114

3Department of Biostatistics, Massachusetts General Hospital, 50 Staniford Street - Suite

560, Boston, MA 02114

Corresponding author:

Miriam A. Bredella

Department of Radiology, Yawkey 6E, 55 Fruit Street

Boston, MA 02114

Phone: 617-726-7717

Fax: 617-726-5282

E-mail: [email protected]

Running head: Body composition predictors of GH deficiency

Articles in PresS. J Appl Physiol (December 18, 2008). doi:10.1152/japplphysiol.90998.2008

Copyright © 2008 by the American Physiological Society.

by 10.220.33.1 on August 16, 2016

http://jap.physiology.org/D

ownloaded from

1

Abstract

Visceral adiposity is a strong determinant of GH secretion and states of GH deficiency

are associated with increased visceral adiposity, and decreased lean body mass. The

purpose of our study was to determine the sensitivity and specificity of different methods

of assessing body composition (anthropometry, DXA, and CT) to predict GH deficiency

in premenopausal women and threshold values for each technique to predict GH

deficiency, using ROC curve analysis. We studied a group of 45 healthy lean,

overweight, and obese premenopausal women who underwent anthropometric

measurements (BMI, waist and hip circumferences, skin fold thickness), DXA, CT and a

GHRH-arginine stimulation test. ROC curve analysis was used to determine cut-off

values for each method to identify GH deficiency. Visceral adiposity measured by CT

showed the highest sensitivity and specificity for identifying subjects with GH deficiency

with a cut-off of >9962 mm2 (AUC: 0.95, sensitivity:100%, specificity:77.8%,

p=0.0001). Largest waist circumference showed high sensitivity and specificity with a

cut-off of >101.7 cm (AUC: 0.89, sensitivity:88.9%, specificity:75%, p=0.0001). When

comparing the ROC curves of visceral fat measured by CT and largest waist

circumference, the difference between the two methods was not statistically significant

(p=0.36). Our study showed that the largest waist circumference predicts the presence of

GH deficiency in healthy premenopausal women with high sensitivity and specificity and

nearly as well as CT measurement of visceral adiposity. It can be used to identify women

in whom GH deficiency is likely and therefore in whom formal GH stimulation testing

might be indicated.

by 10.220.33.1 on August 16, 2016

http://jap.physiology.org/D

ownloaded from

2

Keywords: obesity, body composition, Growth Hormone deficiency, waist

circumference, ROC curve analysis

by 10.220.33.1 on August 16, 2016

http://jap.physiology.org/D

ownloaded from

3

Introduction

Obesity is highly prevalent in the western world, and visceral adiposity is an independent

predictor of metabolic complications such as dyslipidemia, type 2 diabetes and

cardiovascular disease (25, 35). Prior studies have established that visceral adiposity is a

strong determinant of growth hormone (GH) secretion (11, 31, 32), and that GH plays a

role in modulating body composition. States of GH deficiency are associated with

increased body fat, including visceral adiposity, and decreased lean body mass (10, 14);

whereas states of GH excess are associated with decreased body fat and increased lean

body mass (4).

The GHRH-arginine stimulation test is a sensitive and specific test for diagnosing GH

deficiency, and a cut-off limit of 5 ng/ml has been used to diagnose GH deficiency in

adults (6). However, this test is invasive, time-consuming, and expensive. Should GH

deficiency be determined to be a treatable condition in young women with visceral

adiposity, it would be useful to be able to perform a simple body composition

measurement to identify subjects who may be at risk for GH deficiency and for whom

formal GH simulation testing may be more likely to yield a positive result. Computed

tomography (CT) can quantify visceral and subcutaneous fat depots (7) and is the gold

standard for measuring visceral fat. However, it is expensive and involves radiation

exposure. Several clinical methods, including anthropometry, and dual energy X-ray

absorptiometry (DXA), have been used as surrogates for estimating body fat (8, 15, 19,

by 10.220.33.1 on August 16, 2016

http://jap.physiology.org/D

ownloaded from

4

20, 27, 28, 34), but these measurements do not allow for the evaluation of visceral fat

content. The purpose of our study was to compare sensitivity and specificity of simple

less invasive measures of body composition, such as anthropometry to CT and DXA to

predict GH deficiency in a group of lean, overweight, and obese premenopausal women.

In addition, we wanted to determine threshold values for each technique to predict GH

deficiency, using receiver operator characteristic (ROC) curve analysis.

by 10.220.33.1 on August 16, 2016

http://jap.physiology.org/D

ownloaded from

5

Materials and Methods

The study was approved by the institutional review board of Partners Healthcare Inc. and

was Health Insurance Portability and Accountability Act compliant. Written informed

consent was obtained from all subjects prior to the study.

Subjects

The study group comprised 45 healthy premenopausal women who were recruited from

the community through advertisements. Exclusion criteria included hypothalamic or

pituitary disorders, diabetes mellitus or other chronic illnesses, estrogen or glucocorticoid

use and weight greater than 280 pounds (due to the limitations of the DXA and CT

scanners). Participants were admitted to the General Clinical Research Center at the

Massachusetts General Hospital, where testing was performed. Each participant

underwent anthropometric measurements, DXA, and CT, as detailed below, and a

GHRH-arginine stimulation test. For the GHRH-arginine stimulation test, GHRH 1

mcg/kg plus arginine 0.5 g/kg (maximum 30 gm) IV were administered and GH levels

drawn at baseline and every 30 minutes for two hours (6). GH deficiency was based on

standard criteria used to diagnose adults with hypopituitarism (peak GH after stimulation

with GHRH and arginine <5ng/ml) (6). Clinical characteristics, peak GH after GHRH-

arginine stimulation, and fat mass, measured by DXA, have been previously published

(29, 30).

Biochemical Analyses

by 10.220.33.1 on August 16, 2016

http://jap.physiology.org/D

ownloaded from

6

Serum samples were collected and stored at -80˚ C. Serum GH was measured using an

immunoradiometric assay (IRMA) kit, with a minimum detection limit of 0.01 ng/ml, an

intra-assay coefficient of variation (cv) of 3.1-5.4% and an inter-assay cv of 5.9-11.5%.

Anthropometry

Body weight was measured at a standard balance beam scale to the nearest 0.1 kg in

triplicate and averaged. Height was measured barefoot to the nearest 0.1cm in triplicate

and averaged. Body mass index (BMI) was calculated as weight divided by height

squared (kg/m2).

Skin fold thickness of the triceps, biceps, subscapular and suprailiac areas was measured

using metal calipers in triplicate and averaged. Skin fold measurements were used to

estimate % body fat.

Body circumferences were measured at the waist at the smallest circumference between

the lowest rib and iliac crest, at the level of the umbilicus, the midpoint between the

lowest rib and iliac crest, the iliac crest, and at the hip using a metal tape to the nearest

0.1 cm in triplicate and averaged. The largest circumference represents the largest value

obtained from the above measures. Iliac waist-to-hip and largest hip-to-waist ratios were

determined from the circumferential measurements of the waist at the level of the

umbilicus and the hips at the level of the iliac crest taken with the patient in a standing

position.

by 10.220.33.1 on August 16, 2016

http://jap.physiology.org/D

ownloaded from

7

All measurements were performed by research bionutritionists who had been previously

trained and certified to perform these procedures.

Dual X-ray Absorptiometry

DXA measurements of body composition were performed using a Hologic QDR 4500

scanner (Hologic Inc., Waltham, MA). The following parameters were obtained: % fat of

the right and left upper and lower extremities, % trunk fat, and total % body fat. In

addition, the amount of fat (in g) of the upper and lower extremities, trunk, and total body

was obtained. Coefficients of variation of DXA have been reported as 3% for body fat

mass (21).

Computed Tomography (CT)

Each subject underwent cross-sectional CT scan of the abdomen at the level of L4.

Assessment of visceral and subcutaneous abdominal fat compartments by single-slice CT

image of the abdomen was performed. Patients were placed supine, feet first in the

scanner and with flexion of the knees to minimize lumbar lordosis. A lateral scout image

was obtained to identify the level of L4, which served as the landmark for the single slice

image. Scan parameters for each image were standardized (144 table height, 80kV, 70

mA, 2 seconds, 1 cm slice thickness, 48 FOV). Fat attenuation coefficients were set at

-50 to -250 HU as described by Borkan et al. (7). Total abdominal cross-sectional area

was computed by outlining the outer contour of the abdomen. A second outline of the

back and abdominal wall musculature (inner contour) was used to define the

subcutaneous fat area. Visceral abdominal fat was defined as the area within the inner

by 10.220.33.1 on August 16, 2016

http://jap.physiology.org/D

ownloaded from

8

contour comprising all pixels with attenuation coefficients between -50 and -250 HU.

The total fat area was calculated as the sum of subcutaneous fat and visceral abdominal

fat. These values were used to calculate area of subcutaneous adipose tissue, visceral

adipose tissue and total adipose tissue.

Statistical analysis

JMP Statistical Discoveries (version 4.0.2, SAS Institute, Inc., Cary, NC) and MedCalc

(version 9.2.1.0, MedCalc, Mariakerke, Belgium) were used for statistical analysis. The

means and standard deviations (SD) were calculated and groups were compared using the

Student’s t-test. ROC curve analysis of different methods of body composition

measurements was performed to determine sensitivity and specificity, area under the

ROC curve, confidence intervals (CI) as well as cut-off values for each method to detect

GH deficiency.

Since this was an exploratory study we did not perform a validation study. However, we

performed a cross validated error estimate to determine the error estimate for each

measure.

Power calculation: The t-test power was used to approximate the power of the ROC curve

test. With a proposed sample size of 9 for the GH deficient and 36 for the GH sufficient

group, the study will have a power of 82.3% to yield a statistically significant result, that

the area under the ROC curve is greater than 0.5. This computation assumes that the

mean difference is 1.1 (corresponding to means of 1.1 versus 0.0) and the common

within-group standard deviation is 1.0.

by 10.220.33.1 on August 16, 2016

http://jap.physiology.org/D

ownloaded from

9

Results

Clinical Characteristics of Study Subjects

Subject characteristics are shown in Table 1. Study participants ranged from 19-45 years,

mean 33±8.3 years (SD). Study participants ranged in BMI from 19.2 to 43.6 kg/m2,

mean 30.9±6.5 kg/m2 (SD) and were categorized as lean (n=10) if BMI < 25 kg/m2,

overweight (n=12) if BMI ≥ 25 kg/m2 and < 30 kg/m2, and obese (n=23) if BMI ≥30

kg/m2, based on WHO definitions (1). Nine patients had GH deficiency as determined by

the GHRH-arginine stimulation test, and 36 subjects were GH sufficient. Subjects with

GH deficiency were slightly older and had higher weight, BMI, total, subcutaneous and

visceral fat, as determined by CT, compared to the GH sufficient subjects. Clinical

characteristics of the two groups are shown in Table 2.

Body composition determinants of GH deficiency

Results of ROC curve analyses are summarized in Table 3. On the basis of ROC curves,

visceral adiposity measured by CT showed the highest sensitivity and specificity for

identifying subjects with GH deficiency. The area under the curve (AUC) was 0.95 and

with a cut-off value of >9962 mm2, sensitivity was 100% and specificity was 77.8%

(p=0.0001). Sensitivity and specificity of the cross validated error estimate were 89% and

75%, respectively. The largest waist circumference was the umbilical waist

circumference in 75% of patients. Largest waist circumference showed high sensitivity

and specificity when a cut-off value of >101.7 cm was used (AUC: 0.89, sensitivity

88.9%, specificity 75%, p=0.0001). Sensitivity and specificity of the cross validated error

by 10.220.33.1 on August 16, 2016

http://jap.physiology.org/D

ownloaded from

10

estimate were 67% and 72%, respectively. Using a cut-off value of >80cm for largest

waist circumference as used to diagnose metabolic syndrome by the International

Diabetes Federation (3), sensitivity was 100% but specificity was only 9%. Using a cut-

off value of >88cm for the largest waist circumference as proposed by Lean at al (18) to

determine visceral adiposity, sensitivity was 100% but specificity dropped to 25%. When

comparing the ROC curves of visceral fat measured by CT and largest waist

circumference (cut-off value of >101.7cm), the difference between the two methods was

not statistically significant (p=0.36) (Figure 1).

Determination of total and trunk fat content as measured by DXA showed an AUC of

0.87 and 0.88 with sensitivity of 100% and specificity of 69.4% and 66.7% , respectively,

when using a cut-off value of >16246g and >31677g, respectively (p=0.0001). Sensitivity

and specificity of the cross validated error estimate were 67% and 64%, respectively for

total fat and 67% and 67%, respectively, for trunk fat. Comparing ROC curves of visceral

fat measured by CT and trunk fat (g) measured by DXA, the difference between the two

methods was not statistically significant (p=0.36). The largest waist-to-hip ratio

demonstrated an AUC of 0.85 and sensitivity was 100% and specificity was 66.6% when

a cut-off value of >0.85 was used (p= 0.0001). Sensitivity and specificity of the cross

validated error estimate were 67% and 57%, respectively. Comparing the ROC curves of

visceral fat as measured by CT and largest hip to waist ratio, the difference between the

two methods was not statistically significant (p=0.3).

by 10.220.33.1 on August 16, 2016

http://jap.physiology.org/D

ownloaded from

11

Discussion

Our study showed that the largest waist circumference can predict GH deficiency in

premenopausal women and that this measurement is almost as sensitive and specific for

predicting GH deficiency as visceral fat measured by abdominal CT. These data may be

of importance if further research confirms the association of GH deficiency with

increased cardiovascular risk.

An increased prevalence of visceral adiposity and cardiovascular events has been

established in women with GH deficiency due to hypopituitarism. Studies have shown

that decreased GH secretion is an independent risk factor for visceral obesity and

cardiovascular disease in this patient population. Higher cardiovascular mortality in

female GH deficient patients than in males has been found (9, 24). This may reflect a

relatively more severe state of GH deficiency in women compared with men, as GH

secretion is nearly twice as high in young, healthy women as in men (11). Therefore, we

focused our study on healthy overweight and obese women and the relationship between

visceral adiposity and GH deficiency in this patient population.

Multiple studies have demonstrated decreases in visceral adiposity, without a change in

overall weight or BMI in GH deficient patients during physiologic GH administration (5,

15). Although not FDA approved for clinical use, GH replacement has been studied in

subjects with visceral adiposity without pituitary or hypothalamic disease, and may

improve insulin sensitivity over time, potentially due to adipose reduction. In a study by

by 10.220.33.1 on August 16, 2016

http://jap.physiology.org/D

ownloaded from

12

Johannsson et al. (17), administration of low-dose GH to obese men resulted in decreased

visceral fat mass, suggesting a possible therapeutic role for GH in patients with visceral

obesity. In a study evaluating GH administration in obese postmenopausal women (12),

12 months of GH administration reduced the amount of visceral fat and increased thigh

muscle mass, whereas no change in subcutaneous adipose tissue was observed.

Therefore, if further research confirms these effects, it might be useful to develop

diagnostic tools in order to identify subjects with visceral adiposity who might be at risk

for being GH deficient and in whom formal GH stimulation testing should be performed.

Sophisticated imaging modalities such as CT or MRI are able to distinguish visceral from

subcutaneous fat with a high level of precision, but these methods are expensive, time

consuming, and CT involves radiation exposure (2, 26). Simple anthropometric variables

such as waist circumference and waist-to-hip ratio have been used to estimate visceral

adipose tissue. Several studies have indicated that the waist circumference is strongly

related to health risks associated with obesity and that it correlates with visceral fat

measured by CT (13, 16). Our study showed that the largest waist circumference is an

easy and reliable method that can predict GH deficiency in premenopausal women. Waist

circumferences can be measured at several locations. In our study, the waist

circumference measured at the umbilicus corresponded in 75% of study participants to

the largest waist circumference. Visceral adiposity measured with CT showed a higher

sensitivity and specificity in detecting GH deficiency than largest waist circumference.

However, the difference between the two methods was not statistically significant. On the

basis of ROC curves, the most sensitive and specific cut-off was >102 cm in women for

by 10.220.33.1 on August 16, 2016

http://jap.physiology.org/D

ownloaded from

13

largest waist circumference. Current guidelines suggest a cut-off of >88 cm in women on

the basis of detecting many metabolic risk factors (16). In our study, the specificity of

detecting GH deficiency dropped from 75% to 25% when using 88cm as a cut-off. The

International Diabetes Federation suggests a cut-off value for largest waist circumference

of >80 cm to diagnose metabolic syndrome. Using this cut-off value, the specificity in

our study dropped to 9%. In a study by Wahrenberg et al (33) a cut-off of >100 cm was

sensitive and specific for predicting insulin resistance in men and women. Based on our

data, a waist circumference of >102 cm in premenopausal women provides a useful

reference value to identify obese women who may be at risk for GH deficiency and who

should undergo formal GH stimulation testing.

In our study, BMI and iliac waist-to-hip ratio measurements showed low sensitivity and

specificity in predicting GH deficiency. Trunk and total fat as determined by DXA

showed high sensitivity and specificity in detecting GH deficiency. However, sensitivity

and specificity were higher for the largest waist circumference. In addition, DXA requires

radiation exposure. As expected, skin fold measurements did not predict GH deficiency

in our population. We performed skin fold thickness measurements of the triceps, bceps,

subscapular, and suprailiac areas to present the complete spectrum of anthropometric

measurements.

Our study had several limitations. First is the relatively small number of subjects who

were GH deficient (n=9) compared to the GH sufficient subjects (n=36). Second, we only

studied premenopausal women. There are sex- and age-related differences in the relation

of waist measurement to accumulation of visceral adipose tissue (22, 23). Thus, it is

by 10.220.33.1 on August 16, 2016

http://jap.physiology.org/D

ownloaded from

14

likely that different cut-off values for waist circumference as predictors of GH deficiency

would be found in pre and postmenopausal women and in men. We also did not perform

a validation study to test the proposed cut-off values derived in our study in a different

population. However, we performed a cross validated error estimate which confirmed our

cut-off values. Without a larger sample size and without appropriate cross validation data

our cut-off values for waist circumference should be viewed with caution until results of

larger studies have become available. Since there were only 9 cases in our sample,

estimates of sensitivity would have a standard error of 13% (calculated at 80%

sensitivity). A measured sensitivity of 100% would have a 95% lower confidence bound

of 72%. For many clinical applications this is not adequate and further validation studies

in larger patient populations should be performed.

In conclusion, the largest waist circumference predicts the presence of GH deficiency in

premenopausal women without hypothalamic or pituitary disease with high sensitivity

and specificity and nearly as well as CT measurement of visceral adiposity. This provides

further evidence of the importance of visceral fat mass as a predictor and possible

mechanism for GH deficiency in young healthy women. GH replacement has not been

established to be a safe and effective treatment for young overweight or obese women

and is not FDA approved. However, should GH deficiency be established to have

important cardiovascular risk or metabolic consequences in the future, we raise the

possibility that a simple test that can be performed in any office with a tape measure

might be able to identify women in whom GH deficiency is likely and therefore in whom

formal GH stimulation testing might be indicated.

by 10.220.33.1 on August 16, 2016

http://jap.physiology.org/D

ownloaded from

15

Grants

This work was supported in part by the following grants: HL077674, MO1 RR01066, and

K23RR23090

by 10.220.33.1 on August 16, 2016

http://jap.physiology.org/D

ownloaded from

16

References

1. Obesity: preventing and managing the global epidemic. Report of a WHO consultation. World Health Organ Tech Rep Ser 894: i-xii, 1-253, 2000. 2. Abate N, Burns D, Peshock RM, Garg A, and Grundy SM. Estimation of adipose tissue mass by magnetic resonance imaging: validation against dissection in human cadavers. J Lipid Res 35: 1490-1496, 1994. 3. Alberti KG, Zimmet P, and Shaw J. The metabolic syndrome--a new worldwide definition. Lancet 366: 1059-1062, 2005. 4. Bengtsson BA, Brummer RJ, Eden S, and Bosaeus I. Body composition in acromegaly. Clin Endocrinol (Oxf) 30: 121-130, 1989. 5. Bengtsson BA, Eden S, Lonn L, Kvist H, Stokland A, Lindstedt G, Bosaeus I, Tolli J, Sjostrom L, and Isaksson OG. Treatment of adults with growth hormone (GH) deficiency with recombinant human GH. J Clin Endocrinol Metab 76: 309-317, 1993. 6. Biller BM, Samuels MH, Zagar A, Cook DM, Arafah BM, Bonert V, Stavrou S, Kleinberg DL, Chipman JJ, and Hartman ML. Sensitivity and specificity of six tests for the diagnosis of adult GH deficiency. J Clin Endocrinol Metab 87: 2067-2079, 2002. 7. Borkan GA, Gerzof SG, Robbins AH, Hults DE, Silbert CK, and Silbert JE. Assessment of abdominal fat content by computed tomography. Am J Clin Nutr 36: 172-177, 1982. 8. Bosaeus I, Johannsson G, Rosen T, Hallgren P, Tolli J, Sjostrom L, and Bengtsson BA. Comparison of methods to estimate body fat in growth hormone deficient adults. Clin Endocrinol (Oxf) 44: 395-402, 1996. 9. Bulow B, Hagmar L, Mikoczy Z, Nordstrom CH, and Erfurth EM. Increased cerebrovascular mortality in patients with hypopituitarism. Clin Endocrinol (Oxf) 46: 75-81, 1997. 10. Carrel AL and Allen DB. Effects of growth hormone on adipose tissue. J Pediatr Endocrinol Metab 13 Suppl 2: 1003-1009, 2000. 11. Clasey JL, Weltman A, Patrie J, Weltman JY, Pezzoli S, Bouchard C, Thorner MO, and Hartman ML. Abdominal visceral fat and fasting insulin are important predictors of 24-hour GH release independent of age, gender, and other physiological factors. J Clin Endocrinol Metab 86: 3845-3852, 2001. 12. Franco C, Brandberg J, Lonn L, Andersson B, Bengtsson BA, and Johannsson G. Growth hormone treatment reduces abdominal visceral fat in postmenopausal women with abdominal obesity: a 12-month placebo-controlled trial. J Clin Endocrinol Metab 90: 1466-1474, 2005. 13. Freiberg MS, Pencina MJ, D'Agostino RB, Lanier K, Wilson PW, and Vasan RS. BMI vs. waist circumference for identifying vascular risk. Obesity (Silver Spring) 16: 463-469, 2008. 14. Gerver WJ, De Bruin R, Delemarre v d Waal HA, Aldewereld B, Theunissen P, and Westerterp KR. Effects of discontinuation of growth hormone treatment on body composition and metabolism. Horm Res 53: 215-220, 2000. 15. Han TS and Lean ME. Bioelectrical impedance analysis in nutritional research. Nutrition 14: 707-708, 1998.

by 10.220.33.1 on August 16, 2016

http://jap.physiology.org/D

ownloaded from

17

16. Han TS, van Leer EM, Seidell JC, and Lean ME. Waist circumference action levels in the identification of cardiovascular risk factors: prevalence study in a random sample. Bmj 311: 1401-1405, 1995. 17. Johannsson G, Marin P, Lonn L, Ottosson M, Stenlof K, Bjorntorp P, Sjostrom L, and Bengtsson BA. Growth hormone treatment of abdominally obese men reduces abdominal fat mass, improves glucose and lipoprotein metabolism, and reduces diastolic blood pressure. J Clin Endocrinol Metab 82: 727-734, 1997. 18. Lean ME, Han TS, and Morrison CE. Waist circumference as a measure for indicating need for weight management. Bmj 311: 158-161, 1995. 19. Lukaski HC. Methods for the assessment of human body composition: traditional and new. Am J Clin Nutr 46: 537-556, 1987. 20. Mattsson S and Thomas BJ. Development of methods for body composition studies. Phys Med Biol 51: R203-228, 2006. 21. Mazess RB, Barden HS, Bisek JP, and Hanson J. Dual-energy x-ray absorptiometry for total-body and regional bone-mineral and soft-tissue composition. Am J Clin Nutr 51: 1106-1112, 1990. 22. Misra A, Wasir JS, and Vikram NK. Waist circumference criteria for the diagnosis of abdominal obesity are not applicable uniformly to all populations and ethnic groups. Nutrition 21: 969-976, 2005. 23. Molarius A, Seidell JC, Visscher TL, and Hofman A. Misclassification of high-risk older subjects using waist action levels established for young and middle-aged adults--results from the Rotterdam Study. J Am Geriatr Soc 48: 1638-1645, 2000. 24. Nilsson B, Gustavasson-Kadaka E, Bengtsson BA, and Jonsson B. Pituitary adenomas in Sweden between 1958 and 1991: incidence, survival, and mortality. J Clin Endocrinol Metab 85: 1420-1425, 2000. 25. Rexrode KM, Carey VJ, Hennekens CH, Walters EE, Colditz GA, Stampfer MJ, Willett WC, and Manson JE. Abdominal adiposity and coronary heart disease in women. Jama 280: 1843-1848, 1998. 26. Rossner S, Bo WJ, Hiltbrandt E, Hinson W, Karstaedt N, Santago P, Sobol WT, and Crouse JR. Adipose tissue determinations in cadavers--a comparison between cross-sectional planimetry and computed tomography. Int J Obes 14: 893-902, 1990. 27. Smith SR, Lovejoy JC, Greenway F, Ryan D, deJonge L, de la Bretonne J, Volafova J, and Bray GA. Contributions of total body fat, abdominal subcutaneous adipose tissue compartments, and visceral adipose tissue to the metabolic complications of obesity. Metabolism 50: 425-435, 2001. 28. Snijder MB, Visser M, Dekker JM, Seidell JC, Fuerst T, Tylavsky F, Cauley J, Lang T, Nevitt M, and Harris TB. The prediction of visceral fat by dual-energy X-ray absorptiometry in the elderly: a comparison with computed tomography and anthropometry. Int J Obes Relat Metab Disord 26: 984-993, 2002. 29. Utz A, Yamamoto A, Sluss P, Breu J, and Miller KK. Androgens May Mediate a Relative Preservation of IGF-1 Levels in Overweight and Obese Women Despite Reduced Growth Hormone Secretion. J Clin Endocrinol Metab, 2008. 30. Utz AL, Yamamoto A, Hemphill L, and Miller KK. Growth Hormone Deficiency by GHRH/arginine Testing Criteria Predicts Increased Cardiovascular Risk Markers in Normal Young Overweight and Obese Women. J Clin Endocrinol Metab, 2008.

by 10.220.33.1 on August 16, 2016

http://jap.physiology.org/D

ownloaded from

18

31. Vahl N, Jorgensen JO, Jurik AG, and Christiansen JS. Abdominal adiposity and physical fitness are major determinants of the age associated decline in stimulated GH secretion in healthy adults. J Clin Endocrinol Metab 81: 2209-2215, 1996. 32. Vahl N, Jorgensen JO, Skjaerbaek C, Veldhuis JD, Orskov H, and Christiansen JS. Abdominal adiposity rather than age and sex predicts mass and regularity of GH secretion in healthy adults. Am J Physiol 272: E1108-1116, 1997. 33. Wahrenberg H, Hertel K, Leijonhufvud BM, Persson LG, Toft E, and Arner P. Use of waist circumference to predict insulin resistance: retrospective study. Bmj 330: 1363-1364, 2005. 34. Wajchenberg BL. Subcutaneous and visceral adipose tissue: their relation to the metabolic syndrome. Endocr Rev 21: 697-738, 2000. 35. Zhang C, Rexrode KM, van Dam RM, Li TY, and Hu FB. Abdominal obesity and the risk of all-cause, cardiovascular, and cancer mortality: sixteen years of follow-up in US women. Circulation 117: 1658-1667, 2008.

by 10.220.33.1 on August 16, 2016

http://jap.physiology.org/D

ownloaded from

19

Figure legends

Figure 1: ROC curve of visceral fat measured by CT (solid line) and largest waist

circumference (dotted line) to detect GH deficiency. Although the AUC is larger for the

CT measurement, the difference is not significant (AUC CT: 0.95, AUC largest waist

circumference: 0.89, p=0.36).

by 10.220.33.1 on August 16, 2016

http://jap.physiology.org/D

ownloaded from

20

Table legends

Table 1: Clinical characteristics of all subjects.

Table 2: Clinical characteristics of GH deficient and GH sufficient subjects.

Table 3: ROC curve analysis of different body composition methods

by 10.220.33.1 on August 16, 2016

http://jap.physiology.org/D

ownloaded from

by 10.220.33.1 on August 16, 2016

http://jap.physiology.org/D

ownloaded from

Table 1

Clinical characteristics of all subjects.

Variable All subjects

(n= 45)

Age (years) 33±8.3

Weight (kg) 82±18.3

BMI (kg/m2) 30.9±6.5

GH stimulation peak (ng/mL) 14.8±11

by 10.220.33.1 on August 16, 2016

http://jap.physiology.org/D

ownloaded from

Table 2

Clinical characteristics of GH deficient and GH sufficient subjects.

Variable GH peak , 5 ng/ml

(n=9)

GH peak ≥ 5ng/ml

(n=36)

p-value

Age (years) 38.4±6.2 31.7±8.2 0.03

Weight (kg) 101.3±17.3 78.1±15.6 0.0003

BMI (kg/m2) 37±5.2 29.4±6 0.001

Total abdominal fat (mm2)

82078±21911 48026±22625 0.0002

Subcutaneous fat (mm2) 51895±16320 33693±14474 0.002

Visceral fat (mm2) 16852±5349 7127±3868 <0.0001

GH stimulation peak (ng/mL)

3.4±1.2 17.7±10.5 0.0002

by 10.220.33.1 on August 16, 2016

http://jap.physiology.org/D

ownloaded from

Table 3. ROC curve analysis of different body composition methods

Method Threshold Sensitivity Specificity ROC-AUC 95% Confidence Interval p

BMI (kg/m2) >36.4 66.7 88.9 0.83 0.69-0.92 0.0002 Weight (kg) >79.8 100 63.9 0.85 0.71-0.94 0.0001 Waist circumference-largest (cm) >101.7 88.9 75 0.89 0.76-0.96 0.0001 Waist circumference-umbilicus (cm) >101.7 88.9 74.3 0.89 0.75-0.96 0.0001 Waist circumference-iliac (cm) >99 100 66.7 0.84 0.7-0.93 0.0001 Waist circumference-mid (cm) >98.7 88.9 73.5 0.85 0.71-0.94 0.0001 Waist circumference-smallest (cm) >88.7 100 60 0.86 0.72-0.94 0.0001 Hip circumference (cm) >109 100 50 0.8 0.66-0.91 0.0009 Iliac Waist-Hip Ratio >0.85 77.8 58.3 0.7 0.54-0.82 0.057 Largest waist-hip ratio >0.87 100 66.6 0.85 0.72-0.94 0.0001 Skin % body fat >36.4 88.9 68.6 0.79 0.64-0.89 0.002 Skin biceps (mm) >10.7 100 52.8 0.73 0.57-0.85 0.025 Skin subscapular (mm) >18.5 100 33.3 0.64 0.48-0.78 0.18 Skin suprailiac (mm) >22.3 100 47.2 0.71 0.55-0.83 0.043 Skin tricpes (mm) >22.7 100 52.8 0.76 0.61-0.87 0.008 Abd. CT total area (mm2) >66800 88.9 80.6 0.85 0.71-0.94 0.0001 Abd. CT SQ fat (mm2) >54700 66.7 91.7 0.8 0.66-0.91 0.0009 Abd. CT visceral fat (mm2) >9962 100 77.8 0.95 0.84-0.99 0.0001 DXA % fat left arm >43.8 100 61.1 0.8 0.65-0.9 0.0012 DXA % fat right arm >42.1 100 63.9 0.83 0.69-0.92 0.0002 DXA % fat left leg >45 77.8 75 0.77 0.62-0.88 0.0043 DXA % fat right leg >44.6 77.8 77.8 0.76 0.61-0.87 0.0078 DXA % fat trunk >39.9 100 66.7 0.85 0.71-0.94 0.0001 DXA % fat total >39.1 100 63.9 0.83 0.69-0.93 0.0001 DXA fat trunk (g) >16246 100 69.4 0.87 0.74-0.95 0.0001 DXA fat total (g) >31677 100 66.7 0.88 0.74-0.95 0.0001 DXA lean total (g) >45637 100 36.1 0.7 0.55-0.83 0.05 DXA lean trunk (g) >23323 100 61.1 0.77 0.62-0.88 0.004

by 10.220.33.1 on August 16, 2016 http://jap.physiology.org/ Downloaded from

by 10.220.33.1 on August 16, 2016

http://jap.physiology.org/D

ownloaded from