- Email: [email protected]
Body Mass Index Underestimates Adiposity in Persons With Multiple Sclerosis
Accepted Manuscript Body mass index underestimates adiposity in persons with multiple sclerosis Lara A. Pilutti, PhD, Robert W. Motl, PhD PII:
S0003-9993(15)01235-6
DOI:
10.1016/j.apmr.2015.09.014
Reference:
YAPMR 56326
To appear in:
ARCHIVES OF PHYSICAL MEDICINE AND REHABILITATION
Received Date: 30 June 2015 Revised Date:
16 September 2015
Accepted Date: 21 September 2015
Please cite this article as: Pilutti LA, Motl RW, Body mass index underestimates adiposity in persons with multiple sclerosis, ARCHIVES OF PHYSICAL MEDICINE AND REHABILITATION (2015), doi: 10.1016/j.apmr.2015.09.014. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
RUNNING HEAD: Body mass index in multiple sclerosis
RI PT
Title: Body mass index underestimates adiposity in persons with multiple sclerosis
SC
Authors: Lara A. Pilutti, PhD, Robert W. Motl, PhD
From the Department of Kinesiology and Community Health, University of Illinois at Urbana-Champaign,
M AN U
Urbana, Illinois, USA
Acknowledgements:
Disclosure Statement:
TE D
This study was funded, in part, by the National Multiple Sclerosis Society [PP1695; IL 0003; IL 0010].
We certify that no party having a direct interest in the results of the research supporting this
EP
article has or will confer a benefit on us or on any organization with which we are associated and we certify that all financial and material support for this research and work are clearly
AC C
identified in the title page of the manuscript.
Corresponding author:
Please address all correspondence to Lara A. Pilutti, Ph.D., Department of Kinesiology and Community Health, University of Illinois, 213 Freer Hall, Urbana, IL 61801, USA. Phone: 217-333-6126. Fax: 217-244-7322. Electronic Mail: [email protected]
ACCEPTED MANUSCRIPT
For correspondence and reprints, contact Lara A. Pilutti, Ph.D., Department of Kinesiology and
AC C
EP
TE D
M AN U
SC
RI PT
Community Health, University of Illinois, 213 Freer Hall, Urbana, IL 61801, email: [email protected].
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 1 1
Body mass index underestimates adiposity in persons with multiple sclerosis Objective: To examine the relationship between body mass index (BMI) and adiposity
3
determined from dual-energy X-ray absorptiometry (DXA) in persons with multiple sclerosis
4
(MS) and non-MS controls. The accuracy of standard and alternate BMI thresholds for obesity
5
was examined.
6
Design: Cross-sectional.
7
Setting: University research laboratory.
8
Participants: The sample included 235 persons with MS and 53 controls.
9
Interventions: Not applicable.
M AN U
SC
RI PT
2
Main Outcome Measures: Main outcome measures included BMI, whole body soft tissue
11
composition (i.e., %BF, fat mass, and lean soft tissue mass), bone mineral content (BMC), and
12
bone mineral density (BMD).
13
Results: We observed significant, strong associations between BMI and sex-specific %BF in
14
persons with MS and non-MS controls, and BMI explained approximately 40% of the variance in
15
%BF in both the MS and control samples. Receiver operating characteristic (ROC) curve analyses
16
indicated that the standard BMI threshold for obesity (i.e., 30kg/m2) had excellent specificity
17
(93-100%), but poor sensitivity (37-44%) in persons with MS and non-MS controls. The BMI
18
threshold that best identified %BF-defined obesity was 24.7kg/m2 in the MS sample and
19
25.1kg/m2 in the control sample.
20
Conclusions: We determined a strong association between BMI and adiposity; however, the
21
current BMI threshold for classifying obesity underestimates true adiposity in persons with MS.
22
A similar relationship was observed between BMI and obesity in non-MS controls. The non-MS
AC C
EP
TE D
10
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 2 23
sample included primarily middle-age women, and similar BMI-%BF misclassifications have
24
been reported in these samples.
26
RI PT
25
Key words: body mass index; obesity; body composition; multiple sclerosis.
27 28
SC
29 30
M AN U
31 32 33 34
38 39 40 41 42 43 44 45 46
EP
37
AC C
36
TE D
35
ACCEPTED MANUSCRIPT
48
BF
Body fat
49
BMC
Bone mineral content
50
BMD
Bone mineral density
51
BMI
Body mass index
52
DXA
Dual-energy X-ray absorptiometry
53
EDSS
Expanded Disability Status Scale
54
MS
Multiple sclerosis
55
PDDS
Patient Determined Disease Steps
56
ROC
Receiver operating characteristic
57 58
63 64 65 66 67 68 69
EP
62
AC C
61
TE D
59 60
SC
List of Abbreviations
M AN U
47
RI PT
Body mass index in multiple sclerosis 3
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 4
Body mass index (BMI) is a widely used tool for identifying obesity and assessing health risk
71
associated with excess body fat. A BMI of ≥30 kg/m2 has been established by the World Health
72
Organization1 as the threshold for obesity in the general population. Although BMI has many
73
advantages, such as cost-effectiveness and ease of use, there are important limitations of this
74
tool.2–4 In particular, BMI is a surrogate marker of obesity and does not distinguish between fat
75
and lean tissues, and therefore, does not provide a direct measure of an individual’s
76
adiposity.2,4,5 Consequently, BMI might overestimate or underestimate true adiposity, and
77
importantly, lead to a misinterpretation of potential health risk.
M AN U
SC
RI PT
70
78
Previous research has examined the ability of BMI to identify obesity in healthy and clinical
80
populations.2,4,6–8 These studies suggest that a strong association exists between BMI and body
81
fat, but the diagnostic performance of BMI in identifying obesity is limited. There is evidence to
82
suggest that the standard BMI threshold for obesity frequently underestimates adiposity,
83
particularly in older adults and populations with chronic health conditions.2,6–9 For instance, a
84
BMI threshold of 30kg/m2 failed to identify 73.9% of persons who were obese based on percent
85
fat mass from bioelectrical impedance in a sample of 77 individuals with chronic spinal cord
86
injury.7 Similarly, a BMI threshold of 24.9kg/m2, rather than 30kg/m2, demonstrated superior
87
accuracy in identifying obesity based on percent body fat (%BF) from dual-energy X-ray
88
absorptiometry (DXA) in 317 healthy, sedentary, post-menopausal women.6 The development
89
of BMI thresholds that accurately identify true adiposity in healthy and clinical populations has
90
important implications for intervention and management of chronic disease risk.
91
AC C
EP
TE D
79
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 5
There has been increasing interest in the role of adiposity in the development and progression
93
of multiple sclerosis (MS).10,11 Indeed, studies have identified that 50-70% of persons with MS
94
are classified as overweight or obese using the standard BMI thresholds.12–15 If a similar
95
relationship exists between BMI and adiposity in persons with MS as in other clinical
96
populations, the prevalence of obesity, and risk of obesity-related chronic health conditions,
97
would in fact be underestimated. Further, the relationship between BMI and obesity might
98
differ as a function of disability level in persons with MS.8 Indeed, there is evidence for
99
differences in the relationship between BMI and obesity with respect to certain population
M AN U
SC
RI PT
92
characteristics. For instance, the diagnostic performance of BMI for identifying obesity
101
decreases with increasing age2 and is worse in persons with tetraplegia compared to those with
102
paraplegia.16 Examining the diagnostic performance of BMI in identifying obesity in persons
103
with MS is particularly important, as BMI has recently been recommended as one of the core
104
outcome measures for exercise studies as a metric of body composition.17 No studies to date,
105
however, have examined the relationship between BMI and adiposity in MS.
EP
106
TE D
100
To that end, we conducted a secondary analysis of cross-sectional data to examine the
108
relationship between BMI and adiposity assessed by DXA in a large sample of participants with
109
MS and a non-MS reference sample. We sought to determine the accuracy of standard and
110
alternate BMI thresholds for identifying obesity based on sex-specific %BF from DXA. Sex-
111
specific %BF cut points have been commonly used in large-scale epidemiologic research
112
examining the diagnostic performance of BMI in the general adult population.4 We were further
113
interested in examining the influence of MS disability on the relationship between BMI and
AC C
107
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 6
adiposity to provide an indication of potential differences in this relationship with respect to
115
ambulatory ability. Based on the relationship between BMI and obesity reported in older adults
116
and other clinical populations, we expect that BMI will underestimate adiposity in persons with
117
MS. Further, we expect that BMI will have poorer diagnostic performance in persons with MS
118
with greater disability levels. Such an examination will be important for establishing the validity
119
of BMI as a marker of obesity in persons with MS.
METHODS
122
M AN U
121
SC
120
RI PT
114
Participants
123 124
We conducted a secondary analysis of data from 235 persons with MS and 53 non-MS controls
126
who had participated in five previous research studies involving body composition assessment
127
at a University research laboratory between January 2006 and July 2014.18–21 Three studies
128
involved cross-sectional examinations of fitness, functional, and symptomatic outcomes in
129
persons with MS, and two were prospective studies involving physical activity or exercise
130
training interventions. Data from participants with MS was included from five separate (n=77;
131
n=61; n=34; n=33; n=30) research studies and data from control participants was included from
132
two (n=33; n=20) of these studies. The inclusion criteria for all participants were: age 18-65
133
years; ambulatory with or without an assistive device; and absence of risk factors for
134
participation in exercise based on the Physical Activity Readiness Questionnaire.22 Additionally,
135
participants with MS had a clinically definite diagnosis of MS and were relapse-free during the
AC C
EP
TE D
125
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 7 136
past 30 days prior to assessments. Non-MS controls were matched to participants with MS
137
based on sex, age, height, and weight.
139
RI PT
138
Measures
140 141
Disability. The Patient-Determined Disease Steps (PDDS)23 scale was used to characterize the
143
level of neurological disability of the MS sample. The PDDS has demonstrated good validity
144
based on associations with the clinically-determined Expanded Disability Status Scale
145
(EDSS).24,25 To examine the influence of disability on body composition, participants with MS
146
were stratified into two groups based on PDDS scores. Individuals who did not experience gait
147
disability (i.e., PDDS < 3.0) were coded as ‘0’ and participants who did experience gait disability
148
(i.e., PDDS ≥ 3.0) were coded as ‘1’. This would provide an indication of the impact of
149
limitations in walking on body composition. The stratification of PDDS groups based on gait
150
disability has been used in previous research.26,27
151
Height and weight. Height and weight were measured in the laboratory to the nearest 0.1cm or
152
kg, respectively, using a scale-stadiometer.a
153
Body mass index (BMI). BMI was calculated as weight in kilograms divided by height in meters
154
squared. BMI was classified using standard threshold values: underweight, BMI<18.5kg/m2;
155
normal weight, BMI=18.5-24.9kg/m2; overweight, BMI=25.0-29.9kg/m2; and obese,
156
BMI≥30.0kg/m2.1
AC C
EP
TE D
M AN U
SC
142
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 8
Dual-energy X-ray absorptiometry (DXA). Whole body soft tissue composition (i.e., %BF, fat
158
mass, and lean soft tissue mass), bone mineral content (BMC), and bone mineral density (BMD)
159
were assessed by DXA using a Hologic QDR 4500A bone densitometer with Hologic software.b
160
The accuracy of the densitometer was verified daily by scanning the manufacturer’s
161
hydroxyapatite spine phantom of a known density. %BF-defined obesity was determined from
162
DXA and classified as ≥25% BF for men and ≥35% BF for women. These sex-specific %BF
163
thresholds have been used most commonly in large-scale, epidemiologic studies and have been
164
associated with risk factors for cardiometabolic disease, including elevated blood pressure,
165
blood glucose, insulin, cholesterol levels, fibrinogen, and CRP concentrations.2–4,9,28
M AN U
SC
RI PT
157
166 167
Procedures
169
TE D
168
All procedures were reviewed and approved by a University Institutional Review Board.
171
Participants provided written informed consent prior to data collection. Participants visited a
172
university research laboratory to complete demographic, clinical and morphologic assessments.
173
During the testing session, all participants completed a self-report demographic questionnaire
174
and participants with MS further completed the PDDS. Standing height and weight were then
175
assessed by a member of the research team while participants were wearing clothing and
176
footwear. Participants lastly underwent a whole body DXA scan. The scanning protocol was
177
consistent across all studies and conducted using the same Hologic QDR 4500A bone
178
densitometer. All participants wore light weight clothing that was free of metal and removed all
AC C
EP
170
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 9
jewelry prior to scanning. Participants were positioned on the scanner according to the
180
manufacturer’s instructions with the use of positioning aids as necessary. Participants were not
181
scanned in the fasted state.
RI PT
179
182 183
Data analysis
SC
184 185
M AN U
186
Data were analyzed using SPSS version 22.0.c Analyses were conducted using a de-identified
188
dataset. Descriptive statistics were computed to summarize the demographic, clinical, and
189
morphological characteristics of the sample. Values in the text are presented as mean (SD),
190
unless otherwise specified. We examined the differences in demographic and morphologic
191
characteristics between controls, persons with MS overall, and by MS disability level using
192
independent samples t-tests, chi-square tests, and one-way ANOVAs with post-hoc Bonferroni
193
corrections. Bivariate Pearson (r) and Spearman (ρ) correlations were conducted to provide an
194
overall indication of the relationship between BMI and body composition metrics. The
195
magnitude of correlation coefficients was expressed using Cohen’s29 guidelines of .1, .3, and .5
196
as small, moderate, and large, respectively. Fisher’s Z-tests were used to compare the
197
difference in correlation coefficients between controls and the MS sample overall, and between
198
MS disability groups. Linear regression analyses were conducted to determine the contribution
199
of BMI to sex-specific %BF from DXA. The regression equations were used to estimate %BF
200
associated with standard BMI thresholds of 18.5, 25, and 30kg/m2. Receiver-operating
AC C
EP
TE D
187
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 10
characteristic (ROC) analyses were used to examine the accuracy of BMI for classifying obesity
202
using sex-specific %BF-defined obesity obtained from DXA. The overall performance of BMI
203
thresholds was determined by examining sensitivity (i.e., true-positive) and specificity (i.e.,
204
true-negative). The best BMI threshold for identifying %BF-defined obesity was determined as
205
the value that maximized the sum of sensitivity and specificity (i.e., the minimum number of
206
false-positives and false-negatives). Statistical significance was set at p<.05.
207
RESULTS
M AN U
208 209 210
SC
RI PT
201
Participants
211
TE D
212
Participant characteristics are presented in Table 1. There were no significant differences in
214
age, sex, height, or weight between controls and the aggregate MS sample (all p>.05). There
215
were significant differences in MS disability groups in age, PDDS score, disease duration, and
216
disease course (all p<.05). PDDS scores were missing from 8 participants from one study, and
217
therefore were not included in analyses.
219 220 221 222
AC C
218
EP
213
Body composition
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 11
BMI and DXA outcomes for all participants are presented in Table 2. Due to the limited number
224
of participants that were classified as underweight (MS=5; control=1), underweight and normal
225
weight BMI groups were collapsed. Overall, participants with MS had a mean BMI of ~28kg/m2
226
and 29% of the sample was classified as obese using BMI. The overall MS sample had a mean
227
%BF of 35, and 59% of the sample was classified as obese using sex-specific %BF thresholds for
228
obesity. There was a significant difference between groups in BMC (F[2,277]=3.32, p=.04,
229
ηρ2=.02), such that participants with MS with gait disability had a lower BMC than those with
230
MS without gait disability (p=.03). There were no statistically significant differences between
231
groups in any of the other body composition outcomes (all p>.05).
M AN U
SC
RI PT
223
232 233
235
BMI as marker of obesity
TE D
234
Correlation coefficients between BMI and DXA outcomes are presented in Table 3. In the MS
237
sample overall, significant, strong correlations were observed between BMI and %BF. There
238
were no significant differences in the correlation coefficients between non-MS controls and the
239
MS sample overall (all p>.05). Significantly stronger Pearson and Spearman correlation
240
coefficients were observed between BMI and whole body fat mass (p<.001) and BMC (p<.05) in
241
persons with MS with gait disability compared to those with MS without gait disability.
242
Significantly stronger Pearson correlation coefficients were observed between BMI and %BF
243
(p=.03), in persons with MS with gait disability compared to those with MS without gait
244
disability. We further examined the relationship between BMI and %BF by sex, and there were
AC C
EP
236
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 12 245
no significant differences in this relationship between women (r=.69; ρ=.74) and men (r=.72;
246
ρ=.64) (p>.05).
RI PT
247
Linear regression analyses revealed that BMI explained a significant portion of the variance in
249
sex-specific %BF in non-MS controls [F(1,52)=30.13, p< .001, R2=.37] and the MS sample overall
250
[F(1,234)=171.03, p<.001, R2=.42]. The resulting regression line equation was %BF=1.02(BMI) +
251
4.98 for non-MS controls, and %BF=0.83(BMI) + 12.03 for the MS sample overall. Linear
252
regression analyses by disability level revealed that BMI explained a significant portion of the
253
variance in sex-specific %BF in the MS sample without gait disability [F(1,123)=65.04, p<.001,
254
R2=.35] and with gait disability [F(1,102)=130.44, p<.001, R2=.56]. The resulting regression line
255
equation was %BF=0.72(BMI) + 14.72 for participants with MS without gait disability, and
256
%BF=1.09(BMI) + 5.74 for participants with MS with gait disability. The regression equations
257
were used to calculate predicted %BF associated with standard BMI thresholds and are
258
presented in Table 4.
M AN U
TE D
EP
259
SC
248
ROC curves examining the accuracy of BMI for classifying obesity using sex-specific %BF-defined
261
obesity are presented in Figure 1. Overall, the standard BMI threshold for obesity (i.e., 30kg/m2)
262
had excellent specificity, but poor sensitivity in all groups (Figure 1). The BMI threshold that
263
best identified sex-specific %BF-defined obesity was 25.1kg/m2 in the control sample
264
(sensitivity=79%; specificity=81%) and 24.7kg/m2 in the MS sample overall (sensitivity=82%;
265
specificity=77%). The BMI threshold that best identified sex-specific %BF-defined obesity was
AC C
260
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 13 266
26.1kg/m2 in the MS sample without gait disability (sensitivity=76%; specificity=85%) and
267
24.8kg/m2 in the MS sample with gait disability (sensitivity=80%; specificity=87%).
DISCUSSION
269 270
SC
271
RI PT
268
We undertook the first large-scale examination of the relationship between BMI and adiposity
273
in persons with MS. There was a strong association between BMI and sex-specific %BF-
274
determined adiposity, but the primary novel finding was the misclassification of obesity using
275
the traditional BMI threshold (i.e., 30kg/m2), and this was not different from matched, non-MS
276
controls. The current BMI threshold for obesity underestimated true adiposity, resulting in
277
many obese individuals being inaccurately classified as non-obese. Researchers and clinicians
278
should be cautious in the use of traditional BMI thresholds for identifying obesity and assessing
279
health risk in persons with MS.
TE D
EP
280
M AN U
272
We determined that a BMI threshold of approximately 25kg/m2 was a more accurate criterion
282
for defining obesity in persons with MS and matched controls. Similar BMI values for defining
283
obesity have been observed in persons with other neurological disorders,7,16 as well as post-
284
menopausal women.6 The lack of difference in obesity misclassification between persons with
285
MS and non-MS controls is not surprising, and likely attributable to the characteristics of the
286
non-MS sample (i.e., primarily middle-aged women). Previous studies involving samples of post-
287
menopausal women have reported improved accuracy of lower BMI thresholds for defining
AC C
281
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 14
obesity.6 Alternatively, other anthropometric measures of obesity that are easily administered
289
and cost-effective might provide a more accurate estimation of chronic disease risk in persons
290
with MS. For instance, it has been reported that waist circumference (WC) is a better indicator
291
of cardiovascular disease risk than BMI in persons with spinal cord injury.30,31 BMI and WC
292
reflect different aspects of body composition, such that BMI is affected by changes in adipose
293
and lean tissue mass, whereas, WC is primarily affected by changes in central mass.8 Future
294
research should explore the diagnostic performance of WC and other anthropometric measures
295
of obesity in persons with MS.
M AN U
SC
RI PT
288
296
Determining an accurate diagnosis of obesity in persons with MS has critical health implications
298
considering that a substantial number of individuals with MS are overweight or obese according
299
to BMI,8–12 and the widespread use of BMI cut-points for defining obesity. The accumulation of
300
excess body fat has consistently been associated with chronic health conditions,32 and this is
301
particularly important for persons with MS as comorbid health conditions can negatively impact
302
disease diagnosis, disease progression, and health-related quality of life.33 Previous research in
303
a large sample of adults without MS (n=6,123) aged 18-80 years suggests that persons classified
304
as non-obese according to BMI, but obese according to %BF, present with increased
305
cardiometabolic risk factors including high blood pressure, glucose, insulin, triglycerides, low-
306
density lipoprotein, fibrinogen, and C-reactive protein concentrations.3 Using a lower BMI
307
threshold or other anthropometric measures to define obesity could result in the identification
308
of more individuals who are potentially at risk of obesity-related chronic health conditions, and
309
importantly, this would allow for earlier intervention and treatment. It will be important for
AC C
EP
TE D
297
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 15 310
future research to examine the validity of lower BMI thresholds and other anthropometric
311
measures in MS based on cardiometabolic biomarkers, in addition to %BF-define obesity.
RI PT
312
Despite the poor diagnostic performance of BMI cut-points for identifying obesity, we observed
314
significant, strong correlations between BMI and measures of body fat in persons with MS and
315
non-MS controls. BMI was more strongly associated with some body composition outcomes
316
(%BF, fat mass, and BMD) in persons with MS with gait disability compared to those without
317
gait disability. This suggests that BMI might be a better predictor of obesity-related health risk
318
in persons with MS with higher disability levels. In all groups, stronger associations were
319
observed between BMI and measures of body fat than between BMI and lean tissue, suggesting
320
that BMI is specific for adiposity. This is consistent with previous results in a sample of 95
321
patients with coronary artery disease that reported stronger associations between BMI and
322
%BF (ρ=.66) than BMI and lean mass (ρ=.41).9 Considering the relationship between BMI and
323
%BF by sex, we did not determine any differences in these associations. Previous research in
324
the general adult population has reported somewhat stronger associations between BMI and
325
%BF in women (n=7021, ρ=.87) compared to men (n=6580, ρ=.65), although the associations
326
were strong regardless of sex.2 Collectively, these findings support a potentially different
327
association between BMI and body composition in clinical populations, emphasizing the
328
importance of population-specific examinations of this relationship.
AC C
EP
TE D
M AN U
SC
313
329 330
The mechanisms underlying changes in body composition in persons with MS are likely related
331
to processes associated with the disease itself (e.g., chronic inflammation, glucocorticoid usage,
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 16
vitamin D deficiency)10,34,35 as well as physical activity behaviors,36–38 and accordingly, such
333
mechanisms may change with disease and disability progression over time. Indeed, individuals
334
with MS participate in substantially less physical activity than those without MS,36,37 and
335
persons with MS spend approximately 7.5 hours per day sitting.38 Inactivity and excessive
336
sitting likely contribute to shifting in body compartments resulting in the accumulation of
337
adipose tissue and the loss of bone and lean tissue mass. Similar shifts in body composition
338
have been noted with aging,39–41 and in other neurologic populations.42,43 Physical activity levels
339
and sedentary time are associated with disability status in persons with MS,36,38,44–46 and likely
340
contribute to additional changes in body composition with disease progression, although there
341
has yet to be a comprehensive examination of body composition across the disability spectrum
342
in persons with MS. The accumulation of additional adipose tissue and loss of bone and lean
343
mass with disability progression might explain the stronger association observed between BMI
344
and obesity in persons with MS with gait disability compared to those without gait disability in
345
this study.
347
349 350
Study Limitations
AC C
348
EP
346
TE D
M AN U
SC
RI PT
332
351
There were several strengths of this cross-sectional investigation including the large sample,
352
inclusion of a matched, non-MS reference sample, and the assessment of body composition
353
using gold standard measurement (i.e., DXA). This study was limited in that we performed a
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 17
secondary analysis of existing data from several investigations that were not originally designed
355
to examine the relationship between BMI and obesity, although body composition assessments
356
were conducted consistently across all studies. The assessment of body composition was not
357
conducted in a fasted state and this might have confounded our results. There is evidence to
358
suggest consuming a small meal after an overnight fast does not impact body composition
359
assessed by DXA or bioelectrical impedance in older adults,47 although this has not been
360
examined in MS samples. We further acknowledge that the validity of DXA in the evaluation of
361
adipose tissue mass has not been established in persons with MS. Although participants were
362
screened for risk factors for exercise participation, it is possible that participants suffered from
363
body fluid content altering or body fluid shifting diseases, and this could have impacted body
364
composition results. Based on our screening criteria, we might have further excluded potential
365
participants with high levels of body fat and this might have biased our findings. We further
366
acknowledge that we did not control for potential confounding effects of medication or
367
physical activity levels. Finally, disability level was determined by self-report rather than clinical
368
examination, although there is support for the PDDS as a valid patient-reported measure of
369
disability.25 The results of this study are generalizable to individuals with MS who are primarily
370
female, 22-64 years of age (i.e., age range of MS sample overall), have a relapsing disease
371
course, and ambulatory with mild-to-moderate disability.
373 374 375
SC
M AN U
TE D
EP
AC C
372
RI PT
354
CONCLUSIONS
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 18
We provide the first study to comprehensively examine the use of BMI as a marker of obesity in
377
persons with MS. Although a strong association exists between BMI and adiposity, the current
378
BMI threshold for defining obesity underestimates true adiposity. Underestimating true
379
adiposity might result in delayed intervention and treatment, leading to negative health
380
consequences. Researchers and clinicians should be aware of the limitations of BMI as a tool for
381
defining obesity in MS, and should be cautious in the interpretation of current BMI thresholds
382
in this population.
SC
RI PT
376
M AN U
383 384 385 386
390 391 392 393 394 395 396 397
EP
389
AC C
388
TE D
387
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 19
References
398
1.
Report of a WHO Committee. World Health Organ Tech Rep Ser 1995;854:1–452.
400 401
World Health Organization. Physical status: the use and interpretation of anthropometry.
2.
RI PT
399
Romero-Corral A, Somers VK, Sierra-Johnson J, Thomas RJ, Collazo-Clavell ML, Korinek J, et al. Accuracy of body mass index in diagnosing obesity in the adult general population. Int J
403
Obes 2008;32:959–66.
404
3.
SC
402
Gómez-Ambrosi J, Silva C, Galofré JC, Escalada J, Santos S, Millán D, et al. Body mass index classification misses subjects with increased cardiometabolic risk factors related to
406
elevated adiposity. Int J Obes 2012;36:286–94.
407
4.
M AN U
405
Okorodudu DO, Jumean MF, Montori VM, Romero-Corral A, Somers VK, Erwin PJ, et al. Diagnostic performance of body mass index to identify obesity as defined by body
409
adiposity: a systematic review and meta-analysis. Int J Obes 2010;34:791–9.
410
5.
Wellens RI, Roche AF, Khamis HJ, Jackson AS, Pollock ML, Siervogel RM. Relationships between the body mass index and body composition. Obes Res 1996;4:35–44.
411
6.
Blew RM, Sardinha LB, Milliken LA, Teixeira PJ, Going SB, Ferreira DL, et al. Assessing the
EP
412
TE D
408
validity of body mass index standards in early postmenopausal women. Obes Res
414
2002;10:799–808.
415
7.
AC C
413
Laughton GE, Buchholz AC, Martin Ginis KA, Goy RE, SHAPE SCI Research Group. Lowering
416
body mass index cutoffs better identifies obese persons with spinal cord injury. Spinal
417
Cord 2009;47:757–62.
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 20 418
8.
Alschuler KN, Gibbons LE, Rosenberg DE, Ehde DM, Verrall AM, Bamer AM, et al. Body mass index and waist circumference in persons aging with muscular dystrophy, multiple
420
sclerosis, post-polio syndrome, and spinal cord injury. Disabil Health J 2012;5:177–84.
421
9.
RI PT
419
Romero-Corral A, Somers VK, Sierra-Johnson J, Jensen MD, Thomas RJ, Squires RW, et al. Diagnostic performance of body mass index to detect obesity in patients with coronary
423
artery disease. Eur Heart J 2007;28:2087–93.
424
SC
422
10. Hedström AK, Olsson T, Alfredsson L. High body mass index before age 20 is associated with increased risk for multiple sclerosis in both men and women. Mult Scler
426
2012;18:1334–6.
428 429
11. Munger KL, Chitnis T, Ascherio A. Body size and risk of MS in two cohorts of US women. Neurology 2009;73:1543–50.
12. Khurana SR, Bamer AM, Turner AP, Wadhwani RV, Bowen JD, Leipertz SL, et al. The
TE D
427
M AN U
425
430
prevalence of overweight and obesity in veterans with multiple sclerosis. Am J Phys Med
431
Rehabil 2009;88:83–91.
434 435
EP
433
13. Pilutti LA, Dlugonski D, Pula JH, Motl RW. Weight status in persons with multiple sclerosis: implications for mobility outcomes. J Obes 2012;2012:868256. 14. Marrie R, Horwitz R, Cutter G, Tyry T, Campagnolo D, Vollmer T. High frequency of adverse
AC C
432
health behaviors in multiple sclerosis. Mult Scler 2009;15:105–13.
436
15. Slawta JN, Wilcox AR, McCubbin JA, Nalle DJ, Fox SD, Anderson G. Health behaviors, body
437
composition, and coronary heart disease risk in women with multiple sclerosis. Arch Phys
438
Med Rehabil 2003;84:1823–30.
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 21
440 441
16. Yarar-Fisher C, Chen Y, Jackson AB, Hunter GR. Body mass index underestimates adiposity in women with spinal cord injury. Obes 2013;21:1223–5. 17. Paul L, Coote S, Crosbie J, Dixon D, Hale L, Holloway E, et al. Core outcome measures for
RI PT
439
442
exercise studies in people with multiple sclerosis: recommendations from a
443
multidisciplinary consensus meeting. Mult Scler 2014;20:1641–50.
18. Pilutti LA, Dlugonski D, Sandroff BM, Klaren RE, Motl RW. Internet-delivered lifestyle
SC
444
physical activity intervention improves body composition in multiple sclerosis: preliminary
446
evidence from a randomized controlled trial. Arch Phys Med Rehabil 2014;95:1283–8.
447
19. Mojtahedi MC, Snook EM, Motl RW, Evans EM. Bone health in ambulatory individuals with
M AN U
445
448
multiple sclerosis: impact of physical activity, glucocorticoid use, and body composition. J
449
Rehabil Res Dev 2008;45:851–61.
451
20. Ward CL, Suh Y, Lane AD, Yan H, Ranadive SM, Fernhall B, et al. Body composition and
TE D
450
physical function in women with multiple sclerosis. J Rehabil Res Dev 2013;50:1139–47. 21. Snook EM, Mojtahedi MC, Evans EM, McAuley E, Motl RW. Physical activity and body
453
composition among ambulatory individuals with multiple sclerosis. Int J MS Care
454
2005;7:137–42.
456 457 458 459 460
22. Thomas S, Reading J, Shephard RJ. Revision of the Physical Activity Readiness
AC C
455
EP
452
Questionnaire (PAR-Q). Can J Sport Sci 1992;17:338–45. 23. Hohol MJ, Orav EJ, Weiner HL. Disease steps in multiple sclerosis: a simple approach to evaluate disease progression. Neurology 1995;45:251–5. 24. Hadjimichael O, Kerns RD, Rizzo MA, Cutter G, Vollmer T. Persistent pain and uncomfortable sensations in persons with multiple sclerosis. Pain 2007;127:35–41.
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 22 461
25. Learmonth YC, Motl RW, Sandroff BM, Pula JH, Cadavid D. Validation of patient determined disease steps (PDDS) scale scores in persons with multiple sclerosis. BMC
463
Neurol 2013;13:37.
464
RI PT
462
26. Ensari I, Motl RW, McAuley E, Mullen SP, Feinstein A. Patterns and predictors of naturally occurring change in depressive symptoms over a 30-month period in multiple sclerosis.
466
Mult Scler 2014;20:602–9.
468 469
27. Motl RW, Mullen S, McAuley E. Multi-group measurement invariance of the multiple sclerosis walking scale-12? Neurol Res 2012;34:149–52.
M AN U
467
SC
465
28. Peterson MD, Gordon PM, Hurvitz EA. Chronic disease risk among adults with cerebral
470
palsy: the role of premature sarcopoenia, obesity and sedentary behaviour. Obes Rev
471
2013;14:171–82.
473 474
29. Cohen J. Statistical power analysis for the behavioral sciences. 2nd ed. Lawrence Erlbaum
TE D
472
Associates: Hillsdale, NJ, 1988.
30. Buchholz AC, Bugaresti JM. A review of body mass index and waist circumference as markers of obesity and coronary heart disease risk in persons with chronic spinal cord
476
injury. Spinal Cord 2005;43:513–8. 31. Ravensbergen HRJC, Lear SA, Claydon VE. Waist circumference is the best index for
AC C
477
EP
475
478
obesity-related cardiovascular disease risk in individuals with spinal cord injury. J
479
Neurotrauma 2013;31:292–300.
480
32. Haslam DW, James WPT. Obesity. Lancet 2005;366:1197–209.
481
33. Marrie RA, Horwitz RI. Emerging effects of comorbidities on multiple sclerosis. Lancet
482
Neurol 2010;9:820–8.
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 23 483
34. Formica CA, Cosman F, Nieves J, Herbert J, Lindsay R. Reduced bone mass and fat-free mass in women with multiple sclerosis: effects of ambulatory status and glucocorticoid
485
use. Calcif Tissue Int 1997;61:129–33.
RI PT
484
35. Ozgocmen S, Bulut S, Ilhan N, Gulkesen A, Ardicoglu O, Ozkan Y. Vitamin D deficiency and
487
reduced bone mineral density in multiple sclerosis: effect of ambulatory status and
488
functional capacity. J Bone Miner Metab 2005;23:309–13.
490 491 492 493
36. Motl RW, McAuley E, Sandroff BM, Hubbard EA. Descriptive epidemiology of physical activity rates in multiple sclerosis. Acta Neurol Scand 2015;131:422–5.
M AN U
489
SC
486
37. Motl RW, McAuley E, Snook EM. Physical activity and multiple sclerosis: a meta-analysis. Mult Scler 2005;11:459–63.
38. Hubbard EA, Motl RW, Manns PJ. The descriptive epidemiology of daily sitting time as a sedentary behavior in multiple sclerosis. Disabil Health J 2015;
495
doi:10.1016/j.dhjo.2015.06.003.
TE D
494
39. Sowers M, Zheng H, Tomey K, Karvonen-Gutierrez C, Jannausch M, Li X, et al. Changes in
497
body composition in women over six years at midlife: Ovarian and chronological aging. J
498
Clin Endocrinol Metab 2007;92:895–901.
500
40. Guo SS, Zeller C, Chumlea WC, Siervogel RM. Aging, body composition, and lifestyle: the
AC C
499
EP
496
Fels Longitudinal Study. Am J Clin Nutr 1999;70:405–11.
501
41. Fantin F, Francesco VD, Fontana G, Zivelonghi A, Bissoli L, Zoico E, et al. Longitudinal body
502
composition changes in old men and women: Interrelationships with worsening disability.
503
J Gerontol A Biol Sci Med Sci 2007;62:1375–81.
504
42. Kocina DP. Body composition of spinal cord injured adults. Sports Med 1997;23:48–60.
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 24
506 507 508 509
43. Jørgensen L, Jacobsen BK. Changes in muscle mass, fat mass, and bone mineral content in the legs after stroke: a 1 year prospective study. Bone 2001;28:655–9. 44. Motl RW, Goldman M. Physical inactivity, neurological disability, and cardiorespiratory
RI PT
505
fitness in multiple sclerosis. Acta Neurol Scand 2011;123:98–104.
45. Dlugonski D, Pilutti LA, Sandroff BM, Suh Y, Balantrapu S, Motl RW. Steps per day among persons with multiple sclerosis: variation by demographic, clinical, and device
511
characteristics. Arch Phys Med Rehabil 2013;94:1534–9.
46. Hubbard EA, Motl RW. Sedentary behavior is associated with disability status and walking
M AN U
512
SC
510
513
performance, but not cognitive function, in multiple sclerosis. Appl Physiol Nutr Metab
514
2015;40:203–6.
518 519 520 521 522 523 524 525 526
TE D
517
the body composition evaluation of elderly persons. J Nutr Health Aging 2009;13:183–6.
EP
516
47. Vilaça KHC, Ferriolli E, Lima NKC, Paula FJA, Moriguti JC. Effect of fluid and food intake on
AC C
515
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 25 527
Suppliers
529
a. Detecto, 203 E Daugherty St., Webb City, MO 64870.
530
b. Hologic, Inc., 35 Crosby Dr., Bedford, MA 01730.
531
c. SPSS Inc., 233 S Wacker Dr, 11th Fl, Chicago, IL 60606.
SC
532 533
M AN U
534 535 536 537
542 543 544 545 546 547 548
EP
541
AC C
540
TE D
538 539
RI PT
528
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 26 549
Figure Legend
550
Fig 1 Receiver-operating characteristic (ROC) curves for BMI to detect sex-specific %BF-define
552
obesity for non-MS controls, participants with MS overall, and by level of disability. (a) non-MS
553
controls: AUC=.833, specificity=94%, sensitivity=37%; (b) MS sample overall: AUC=.846,
554
specificity=93%, sensitivity=44%; (c) PDDS<3.0: AUC=.815, specificity=93%, sensitivity=44%; and
555
(d) PDDS≥3.0: AUC=.913, specificity=100%, sensitivity=44%.
AC C
EP
TE D
M AN U
SC
RI PT
551
ACCEPTED MANUSCRIPT
Table 1 Demographic and clinical characteristics of non-MS controls, and participants with MS overall and by level of disability. MS
P value
Overall
PDSS <3.0
PDDS ≥3.0
Control vs
PDDS
(n=53)
(n=235)
(n=124)
(n=103)
MS Overall
groups
49.2 (11.1)
48.1 (9.3)
45.7 (9.8)
50.8 (8.0)
.45
<.001
43/10
181/54
96/28
78/25
.52
.76
Height, cm
166.6 (9.4)
167.8 (8.9)
168.1 (9.1)
167.2 (8.8)
.40
.49
Weight, kg
73.0 (17.3)
77.8 (19.4)
78.5 (18.6)
76.3 (19.3)
.10
.38
PDDS, mdn (IQR)*
N/A
2.0 (3.0)
1.0 (1.0)
4.0 (2.0)
N/A
<.001
Disease duration, years*
N/A
10.8 (7.9)
9.2 (7.2)
12.9 (8.5)
N/A
<.001
Disease course,
N/A
194/41
117/7
70/33
N/A
<.001
Sex, female/male
TE D
relapsing/progressive*
SC
Age, years*
M AN U
Characteristic
RI PT
Control
BMI=body mass index. IQR=interquartile range. mdn=median. Values are mean (SD), unless otherwise
AC C
EP
noted. *Statistically significant difference between MS disability groups (i.e., PDDS <3.0 vs. PDDS ≥3.0).
ACCEPTED MANUSCRIPT
Table 2 Morphological characteristics of non-MS controls, and participants with MS overall and by level of disability.
Control (n=53)
Overall (n=235)
PDSS <3.0 (n=124)
PDDS ≥3.0 (n=103)
26.4 (6.5)
27.7 (6.9)
27.9 (7.1)
27.2 (6.3)
Under-normal, n (%)
27 (50.9%)
105 (44.7%)
SC
Characteristic
RI PT
MS
46 (44.7%)
Overweight, n (%)
15 (28.3%)
61 (26.0%)
32 (27.2%)
28 (27.2%)
Obese, n (%)
11 (20.8%)
69 (29.0%)
37 (29.8%)
29 (28.2%)
31.9 (10.8)
35.0 (8.8)
34.7 (8.6)
35.3 (9.2)
27 (50.9%)
139 (59.1%)
71 (57.3%)
65 (63.1%)
28,260.1 (12,943.0)
28,359.4 (12,289.5)
28,004.2 (12,913.2)
47,175.4 (9,534.7)
48,068.2 (9,194.1)
46,175.8 (9,489.8)
2,299.6 (350.1)
2,358.7 (344.7)
2,234.5 (344.4)
1.122 (0.092)
1.137 (0.095)
1.106 (0.087)
BMI, kg/m2
Obese, n (%)
24,6334.0 (12,607.6)
Whole body LST mass, g
48,019.6 (10,310.5)
Whole body BMD, g/cm2
2,318.9 (438.5)
AC C
Whole body BMC, g*
EP
Whole body fat mass, g
TE D
Whole body fat, (%)
55 (44.4%)
M AN U
BMI classification
1.124 (0.121)
BMC= bone mineral content. BMD= bone mineral density. BMI=body mass index. LST= lean soft tissue. Values are mean (SD), unless otherwise noted. *Statistically significant difference between MS disability groups (i.e., PDDS <3.0 vs. PDDS ≥3.0).
ACCEPTED MANUSCRIPT
Table 3 Bivariate Pearson and Spearman correlation coefficients between BMI and DXA outcomes in non-MS controls, and participants with MS
RI PT
overall and by level of disability. MS
PDDS ≥3.0
(n=235)
(n=124)
(n=103)
Control (n=53)
Percent body fat^
.61* (.41-.76)
.65* (.46-.78) Whole body fat mass, g#
.75* (.65-.82)
.83* (.79-.87)
.76* (.67-.83)
.92* (.88-.95)
.80* (.75-.84)
.72* (.62-.80)
.92* (.88-.95)
.41* (.16-.61)
.46* (.35-.56)
.40* (.24-.54)
.52* (.36-.65)
.39* (.13-.60)
.48* (.38-.57)
.44* (.29-.57)
.56* (.41-.68)
.34* (.08-.56)
.23* (.11-.35)
.10 (-.08-.27)
.39* (.21-.54)
.27* (.15-.39)
.13 (-.05-.30)
.42* (.25-.57)
.37* (.11-.58)
.23* (.11-.35)
.25* (.08-.41)
.22* (.03-.40)
.46* (.22-.65)
.21* (.08-.33)
.17 (-.01-.34)
.24* (.05-.41)
TE D
EP AC C
.42* (.17-.62)
Whole body bone mineral density, g/cm2
.75* (.65-.82)
.61* (.49-.71)
.80* (.68-.88)
Whole body bone mineral content, g#
.59* (.46-.69)
.66* (.58-.73)
.89* (.82-.94)
Whole body lean soft tissue mass, g
.65* (.57-.72)
SC
PDSS <3.0
M AN U
Characteristic
Overall
ACCEPTED MANUSCRIPT
2 Spearman correlation coefficients are presented below Pearson. 95% confidence intervals are presented in brackets. *Statistically significant
RI PT
correlation coefficient. #Statistically significant difference in Pearson and Spearman correlation coefficients between MS disability groups (i.e., PDDS <3.0 vs. PDDS ≥3.0). ^Statistically significant difference in Pearson correlation coefficients between MS disability groups (i.e., PDDS <3.0 vs.
AC C
EP
TE D
M AN U
SC
PDDS ≥3.0).
ACCEPTED MANUSCRIPT
Table 4 Predicted %BF associated with standard BMI thresholds based on regression equations in non-MS controls, and participants with MS
RI PT
overall and by level of disability.
Overall
PDSS <3.0
PDDS ≥3.0
(kg/m2)
(n=53)
(n=235)
(n=124)
(n=103)
18.5
23.9
27.4
28.0
25.9
25
30.2
32.8
32.7
33.0
30
35.6
36.9
36.3
38.4
M AN U
Control
AC C
EP
TE D
BMI threshold
SC
MS
ACCEPTED MANUSCRIPT
a
M AN U
SC
RI PT
b
d
AC C
EP
TE D
c
S0003-9993(15)01235-6
DOI:
10.1016/j.apmr.2015.09.014
Reference:
YAPMR 56326
To appear in:
ARCHIVES OF PHYSICAL MEDICINE AND REHABILITATION
Received Date: 30 June 2015 Revised Date:
16 September 2015
Accepted Date: 21 September 2015
Please cite this article as: Pilutti LA, Motl RW, Body mass index underestimates adiposity in persons with multiple sclerosis, ARCHIVES OF PHYSICAL MEDICINE AND REHABILITATION (2015), doi: 10.1016/j.apmr.2015.09.014. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
RUNNING HEAD: Body mass index in multiple sclerosis
RI PT
Title: Body mass index underestimates adiposity in persons with multiple sclerosis
SC
Authors: Lara A. Pilutti, PhD, Robert W. Motl, PhD
From the Department of Kinesiology and Community Health, University of Illinois at Urbana-Champaign,
M AN U
Urbana, Illinois, USA
Acknowledgements:
Disclosure Statement:
TE D
This study was funded, in part, by the National Multiple Sclerosis Society [PP1695; IL 0003; IL 0010].
We certify that no party having a direct interest in the results of the research supporting this
EP
article has or will confer a benefit on us or on any organization with which we are associated and we certify that all financial and material support for this research and work are clearly
AC C
identified in the title page of the manuscript.
Corresponding author:
Please address all correspondence to Lara A. Pilutti, Ph.D., Department of Kinesiology and Community Health, University of Illinois, 213 Freer Hall, Urbana, IL 61801, USA. Phone: 217-333-6126. Fax: 217-244-7322. Electronic Mail: [email protected]
ACCEPTED MANUSCRIPT
For correspondence and reprints, contact Lara A. Pilutti, Ph.D., Department of Kinesiology and
AC C
EP
TE D
M AN U
SC
RI PT
Community Health, University of Illinois, 213 Freer Hall, Urbana, IL 61801, email: [email protected].
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 1 1
Body mass index underestimates adiposity in persons with multiple sclerosis Objective: To examine the relationship between body mass index (BMI) and adiposity
3
determined from dual-energy X-ray absorptiometry (DXA) in persons with multiple sclerosis
4
(MS) and non-MS controls. The accuracy of standard and alternate BMI thresholds for obesity
5
was examined.
6
Design: Cross-sectional.
7
Setting: University research laboratory.
8
Participants: The sample included 235 persons with MS and 53 controls.
9
Interventions: Not applicable.
M AN U
SC
RI PT
2
Main Outcome Measures: Main outcome measures included BMI, whole body soft tissue
11
composition (i.e., %BF, fat mass, and lean soft tissue mass), bone mineral content (BMC), and
12
bone mineral density (BMD).
13
Results: We observed significant, strong associations between BMI and sex-specific %BF in
14
persons with MS and non-MS controls, and BMI explained approximately 40% of the variance in
15
%BF in both the MS and control samples. Receiver operating characteristic (ROC) curve analyses
16
indicated that the standard BMI threshold for obesity (i.e., 30kg/m2) had excellent specificity
17
(93-100%), but poor sensitivity (37-44%) in persons with MS and non-MS controls. The BMI
18
threshold that best identified %BF-defined obesity was 24.7kg/m2 in the MS sample and
19
25.1kg/m2 in the control sample.
20
Conclusions: We determined a strong association between BMI and adiposity; however, the
21
current BMI threshold for classifying obesity underestimates true adiposity in persons with MS.
22
A similar relationship was observed between BMI and obesity in non-MS controls. The non-MS
AC C
EP
TE D
10
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 2 23
sample included primarily middle-age women, and similar BMI-%BF misclassifications have
24
been reported in these samples.
26
RI PT
25
Key words: body mass index; obesity; body composition; multiple sclerosis.
27 28
SC
29 30
M AN U
31 32 33 34
38 39 40 41 42 43 44 45 46
EP
37
AC C
36
TE D
35
ACCEPTED MANUSCRIPT
48
BF
Body fat
49
BMC
Bone mineral content
50
BMD
Bone mineral density
51
BMI
Body mass index
52
DXA
Dual-energy X-ray absorptiometry
53
EDSS
Expanded Disability Status Scale
54
MS
Multiple sclerosis
55
PDDS
Patient Determined Disease Steps
56
ROC
Receiver operating characteristic
57 58
63 64 65 66 67 68 69
EP
62
AC C
61
TE D
59 60
SC
List of Abbreviations
M AN U
47
RI PT
Body mass index in multiple sclerosis 3
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 4
Body mass index (BMI) is a widely used tool for identifying obesity and assessing health risk
71
associated with excess body fat. A BMI of ≥30 kg/m2 has been established by the World Health
72
Organization1 as the threshold for obesity in the general population. Although BMI has many
73
advantages, such as cost-effectiveness and ease of use, there are important limitations of this
74
tool.2–4 In particular, BMI is a surrogate marker of obesity and does not distinguish between fat
75
and lean tissues, and therefore, does not provide a direct measure of an individual’s
76
adiposity.2,4,5 Consequently, BMI might overestimate or underestimate true adiposity, and
77
importantly, lead to a misinterpretation of potential health risk.
M AN U
SC
RI PT
70
78
Previous research has examined the ability of BMI to identify obesity in healthy and clinical
80
populations.2,4,6–8 These studies suggest that a strong association exists between BMI and body
81
fat, but the diagnostic performance of BMI in identifying obesity is limited. There is evidence to
82
suggest that the standard BMI threshold for obesity frequently underestimates adiposity,
83
particularly in older adults and populations with chronic health conditions.2,6–9 For instance, a
84
BMI threshold of 30kg/m2 failed to identify 73.9% of persons who were obese based on percent
85
fat mass from bioelectrical impedance in a sample of 77 individuals with chronic spinal cord
86
injury.7 Similarly, a BMI threshold of 24.9kg/m2, rather than 30kg/m2, demonstrated superior
87
accuracy in identifying obesity based on percent body fat (%BF) from dual-energy X-ray
88
absorptiometry (DXA) in 317 healthy, sedentary, post-menopausal women.6 The development
89
of BMI thresholds that accurately identify true adiposity in healthy and clinical populations has
90
important implications for intervention and management of chronic disease risk.
91
AC C
EP
TE D
79
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 5
There has been increasing interest in the role of adiposity in the development and progression
93
of multiple sclerosis (MS).10,11 Indeed, studies have identified that 50-70% of persons with MS
94
are classified as overweight or obese using the standard BMI thresholds.12–15 If a similar
95
relationship exists between BMI and adiposity in persons with MS as in other clinical
96
populations, the prevalence of obesity, and risk of obesity-related chronic health conditions,
97
would in fact be underestimated. Further, the relationship between BMI and obesity might
98
differ as a function of disability level in persons with MS.8 Indeed, there is evidence for
99
differences in the relationship between BMI and obesity with respect to certain population
M AN U
SC
RI PT
92
characteristics. For instance, the diagnostic performance of BMI for identifying obesity
101
decreases with increasing age2 and is worse in persons with tetraplegia compared to those with
102
paraplegia.16 Examining the diagnostic performance of BMI in identifying obesity in persons
103
with MS is particularly important, as BMI has recently been recommended as one of the core
104
outcome measures for exercise studies as a metric of body composition.17 No studies to date,
105
however, have examined the relationship between BMI and adiposity in MS.
EP
106
TE D
100
To that end, we conducted a secondary analysis of cross-sectional data to examine the
108
relationship between BMI and adiposity assessed by DXA in a large sample of participants with
109
MS and a non-MS reference sample. We sought to determine the accuracy of standard and
110
alternate BMI thresholds for identifying obesity based on sex-specific %BF from DXA. Sex-
111
specific %BF cut points have been commonly used in large-scale epidemiologic research
112
examining the diagnostic performance of BMI in the general adult population.4 We were further
113
interested in examining the influence of MS disability on the relationship between BMI and
AC C
107
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 6
adiposity to provide an indication of potential differences in this relationship with respect to
115
ambulatory ability. Based on the relationship between BMI and obesity reported in older adults
116
and other clinical populations, we expect that BMI will underestimate adiposity in persons with
117
MS. Further, we expect that BMI will have poorer diagnostic performance in persons with MS
118
with greater disability levels. Such an examination will be important for establishing the validity
119
of BMI as a marker of obesity in persons with MS.
METHODS
122
M AN U
121
SC
120
RI PT
114
Participants
123 124
We conducted a secondary analysis of data from 235 persons with MS and 53 non-MS controls
126
who had participated in five previous research studies involving body composition assessment
127
at a University research laboratory between January 2006 and July 2014.18–21 Three studies
128
involved cross-sectional examinations of fitness, functional, and symptomatic outcomes in
129
persons with MS, and two were prospective studies involving physical activity or exercise
130
training interventions. Data from participants with MS was included from five separate (n=77;
131
n=61; n=34; n=33; n=30) research studies and data from control participants was included from
132
two (n=33; n=20) of these studies. The inclusion criteria for all participants were: age 18-65
133
years; ambulatory with or without an assistive device; and absence of risk factors for
134
participation in exercise based on the Physical Activity Readiness Questionnaire.22 Additionally,
135
participants with MS had a clinically definite diagnosis of MS and were relapse-free during the
AC C
EP
TE D
125
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 7 136
past 30 days prior to assessments. Non-MS controls were matched to participants with MS
137
based on sex, age, height, and weight.
139
RI PT
138
Measures
140 141
Disability. The Patient-Determined Disease Steps (PDDS)23 scale was used to characterize the
143
level of neurological disability of the MS sample. The PDDS has demonstrated good validity
144
based on associations with the clinically-determined Expanded Disability Status Scale
145
(EDSS).24,25 To examine the influence of disability on body composition, participants with MS
146
were stratified into two groups based on PDDS scores. Individuals who did not experience gait
147
disability (i.e., PDDS < 3.0) were coded as ‘0’ and participants who did experience gait disability
148
(i.e., PDDS ≥ 3.0) were coded as ‘1’. This would provide an indication of the impact of
149
limitations in walking on body composition. The stratification of PDDS groups based on gait
150
disability has been used in previous research.26,27
151
Height and weight. Height and weight were measured in the laboratory to the nearest 0.1cm or
152
kg, respectively, using a scale-stadiometer.a
153
Body mass index (BMI). BMI was calculated as weight in kilograms divided by height in meters
154
squared. BMI was classified using standard threshold values: underweight, BMI<18.5kg/m2;
155
normal weight, BMI=18.5-24.9kg/m2; overweight, BMI=25.0-29.9kg/m2; and obese,
156
BMI≥30.0kg/m2.1
AC C
EP
TE D
M AN U
SC
142
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 8
Dual-energy X-ray absorptiometry (DXA). Whole body soft tissue composition (i.e., %BF, fat
158
mass, and lean soft tissue mass), bone mineral content (BMC), and bone mineral density (BMD)
159
were assessed by DXA using a Hologic QDR 4500A bone densitometer with Hologic software.b
160
The accuracy of the densitometer was verified daily by scanning the manufacturer’s
161
hydroxyapatite spine phantom of a known density. %BF-defined obesity was determined from
162
DXA and classified as ≥25% BF for men and ≥35% BF for women. These sex-specific %BF
163
thresholds have been used most commonly in large-scale, epidemiologic studies and have been
164
associated with risk factors for cardiometabolic disease, including elevated blood pressure,
165
blood glucose, insulin, cholesterol levels, fibrinogen, and CRP concentrations.2–4,9,28
M AN U
SC
RI PT
157
166 167
Procedures
169
TE D
168
All procedures were reviewed and approved by a University Institutional Review Board.
171
Participants provided written informed consent prior to data collection. Participants visited a
172
university research laboratory to complete demographic, clinical and morphologic assessments.
173
During the testing session, all participants completed a self-report demographic questionnaire
174
and participants with MS further completed the PDDS. Standing height and weight were then
175
assessed by a member of the research team while participants were wearing clothing and
176
footwear. Participants lastly underwent a whole body DXA scan. The scanning protocol was
177
consistent across all studies and conducted using the same Hologic QDR 4500A bone
178
densitometer. All participants wore light weight clothing that was free of metal and removed all
AC C
EP
170
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 9
jewelry prior to scanning. Participants were positioned on the scanner according to the
180
manufacturer’s instructions with the use of positioning aids as necessary. Participants were not
181
scanned in the fasted state.
RI PT
179
182 183
Data analysis
SC
184 185
M AN U
186
Data were analyzed using SPSS version 22.0.c Analyses were conducted using a de-identified
188
dataset. Descriptive statistics were computed to summarize the demographic, clinical, and
189
morphological characteristics of the sample. Values in the text are presented as mean (SD),
190
unless otherwise specified. We examined the differences in demographic and morphologic
191
characteristics between controls, persons with MS overall, and by MS disability level using
192
independent samples t-tests, chi-square tests, and one-way ANOVAs with post-hoc Bonferroni
193
corrections. Bivariate Pearson (r) and Spearman (ρ) correlations were conducted to provide an
194
overall indication of the relationship between BMI and body composition metrics. The
195
magnitude of correlation coefficients was expressed using Cohen’s29 guidelines of .1, .3, and .5
196
as small, moderate, and large, respectively. Fisher’s Z-tests were used to compare the
197
difference in correlation coefficients between controls and the MS sample overall, and between
198
MS disability groups. Linear regression analyses were conducted to determine the contribution
199
of BMI to sex-specific %BF from DXA. The regression equations were used to estimate %BF
200
associated with standard BMI thresholds of 18.5, 25, and 30kg/m2. Receiver-operating
AC C
EP
TE D
187
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 10
characteristic (ROC) analyses were used to examine the accuracy of BMI for classifying obesity
202
using sex-specific %BF-defined obesity obtained from DXA. The overall performance of BMI
203
thresholds was determined by examining sensitivity (i.e., true-positive) and specificity (i.e.,
204
true-negative). The best BMI threshold for identifying %BF-defined obesity was determined as
205
the value that maximized the sum of sensitivity and specificity (i.e., the minimum number of
206
false-positives and false-negatives). Statistical significance was set at p<.05.
207
RESULTS
M AN U
208 209 210
SC
RI PT
201
Participants
211
TE D
212
Participant characteristics are presented in Table 1. There were no significant differences in
214
age, sex, height, or weight between controls and the aggregate MS sample (all p>.05). There
215
were significant differences in MS disability groups in age, PDDS score, disease duration, and
216
disease course (all p<.05). PDDS scores were missing from 8 participants from one study, and
217
therefore were not included in analyses.
219 220 221 222
AC C
218
EP
213
Body composition
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 11
BMI and DXA outcomes for all participants are presented in Table 2. Due to the limited number
224
of participants that were classified as underweight (MS=5; control=1), underweight and normal
225
weight BMI groups were collapsed. Overall, participants with MS had a mean BMI of ~28kg/m2
226
and 29% of the sample was classified as obese using BMI. The overall MS sample had a mean
227
%BF of 35, and 59% of the sample was classified as obese using sex-specific %BF thresholds for
228
obesity. There was a significant difference between groups in BMC (F[2,277]=3.32, p=.04,
229
ηρ2=.02), such that participants with MS with gait disability had a lower BMC than those with
230
MS without gait disability (p=.03). There were no statistically significant differences between
231
groups in any of the other body composition outcomes (all p>.05).
M AN U
SC
RI PT
223
232 233
235
BMI as marker of obesity
TE D
234
Correlation coefficients between BMI and DXA outcomes are presented in Table 3. In the MS
237
sample overall, significant, strong correlations were observed between BMI and %BF. There
238
were no significant differences in the correlation coefficients between non-MS controls and the
239
MS sample overall (all p>.05). Significantly stronger Pearson and Spearman correlation
240
coefficients were observed between BMI and whole body fat mass (p<.001) and BMC (p<.05) in
241
persons with MS with gait disability compared to those with MS without gait disability.
242
Significantly stronger Pearson correlation coefficients were observed between BMI and %BF
243
(p=.03), in persons with MS with gait disability compared to those with MS without gait
244
disability. We further examined the relationship between BMI and %BF by sex, and there were
AC C
EP
236
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 12 245
no significant differences in this relationship between women (r=.69; ρ=.74) and men (r=.72;
246
ρ=.64) (p>.05).
RI PT
247
Linear regression analyses revealed that BMI explained a significant portion of the variance in
249
sex-specific %BF in non-MS controls [F(1,52)=30.13, p< .001, R2=.37] and the MS sample overall
250
[F(1,234)=171.03, p<.001, R2=.42]. The resulting regression line equation was %BF=1.02(BMI) +
251
4.98 for non-MS controls, and %BF=0.83(BMI) + 12.03 for the MS sample overall. Linear
252
regression analyses by disability level revealed that BMI explained a significant portion of the
253
variance in sex-specific %BF in the MS sample without gait disability [F(1,123)=65.04, p<.001,
254
R2=.35] and with gait disability [F(1,102)=130.44, p<.001, R2=.56]. The resulting regression line
255
equation was %BF=0.72(BMI) + 14.72 for participants with MS without gait disability, and
256
%BF=1.09(BMI) + 5.74 for participants with MS with gait disability. The regression equations
257
were used to calculate predicted %BF associated with standard BMI thresholds and are
258
presented in Table 4.
M AN U
TE D
EP
259
SC
248
ROC curves examining the accuracy of BMI for classifying obesity using sex-specific %BF-defined
261
obesity are presented in Figure 1. Overall, the standard BMI threshold for obesity (i.e., 30kg/m2)
262
had excellent specificity, but poor sensitivity in all groups (Figure 1). The BMI threshold that
263
best identified sex-specific %BF-defined obesity was 25.1kg/m2 in the control sample
264
(sensitivity=79%; specificity=81%) and 24.7kg/m2 in the MS sample overall (sensitivity=82%;
265
specificity=77%). The BMI threshold that best identified sex-specific %BF-defined obesity was
AC C
260
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 13 266
26.1kg/m2 in the MS sample without gait disability (sensitivity=76%; specificity=85%) and
267
24.8kg/m2 in the MS sample with gait disability (sensitivity=80%; specificity=87%).
DISCUSSION
269 270
SC
271
RI PT
268
We undertook the first large-scale examination of the relationship between BMI and adiposity
273
in persons with MS. There was a strong association between BMI and sex-specific %BF-
274
determined adiposity, but the primary novel finding was the misclassification of obesity using
275
the traditional BMI threshold (i.e., 30kg/m2), and this was not different from matched, non-MS
276
controls. The current BMI threshold for obesity underestimated true adiposity, resulting in
277
many obese individuals being inaccurately classified as non-obese. Researchers and clinicians
278
should be cautious in the use of traditional BMI thresholds for identifying obesity and assessing
279
health risk in persons with MS.
TE D
EP
280
M AN U
272
We determined that a BMI threshold of approximately 25kg/m2 was a more accurate criterion
282
for defining obesity in persons with MS and matched controls. Similar BMI values for defining
283
obesity have been observed in persons with other neurological disorders,7,16 as well as post-
284
menopausal women.6 The lack of difference in obesity misclassification between persons with
285
MS and non-MS controls is not surprising, and likely attributable to the characteristics of the
286
non-MS sample (i.e., primarily middle-aged women). Previous studies involving samples of post-
287
menopausal women have reported improved accuracy of lower BMI thresholds for defining
AC C
281
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 14
obesity.6 Alternatively, other anthropometric measures of obesity that are easily administered
289
and cost-effective might provide a more accurate estimation of chronic disease risk in persons
290
with MS. For instance, it has been reported that waist circumference (WC) is a better indicator
291
of cardiovascular disease risk than BMI in persons with spinal cord injury.30,31 BMI and WC
292
reflect different aspects of body composition, such that BMI is affected by changes in adipose
293
and lean tissue mass, whereas, WC is primarily affected by changes in central mass.8 Future
294
research should explore the diagnostic performance of WC and other anthropometric measures
295
of obesity in persons with MS.
M AN U
SC
RI PT
288
296
Determining an accurate diagnosis of obesity in persons with MS has critical health implications
298
considering that a substantial number of individuals with MS are overweight or obese according
299
to BMI,8–12 and the widespread use of BMI cut-points for defining obesity. The accumulation of
300
excess body fat has consistently been associated with chronic health conditions,32 and this is
301
particularly important for persons with MS as comorbid health conditions can negatively impact
302
disease diagnosis, disease progression, and health-related quality of life.33 Previous research in
303
a large sample of adults without MS (n=6,123) aged 18-80 years suggests that persons classified
304
as non-obese according to BMI, but obese according to %BF, present with increased
305
cardiometabolic risk factors including high blood pressure, glucose, insulin, triglycerides, low-
306
density lipoprotein, fibrinogen, and C-reactive protein concentrations.3 Using a lower BMI
307
threshold or other anthropometric measures to define obesity could result in the identification
308
of more individuals who are potentially at risk of obesity-related chronic health conditions, and
309
importantly, this would allow for earlier intervention and treatment. It will be important for
AC C
EP
TE D
297
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 15 310
future research to examine the validity of lower BMI thresholds and other anthropometric
311
measures in MS based on cardiometabolic biomarkers, in addition to %BF-define obesity.
RI PT
312
Despite the poor diagnostic performance of BMI cut-points for identifying obesity, we observed
314
significant, strong correlations between BMI and measures of body fat in persons with MS and
315
non-MS controls. BMI was more strongly associated with some body composition outcomes
316
(%BF, fat mass, and BMD) in persons with MS with gait disability compared to those without
317
gait disability. This suggests that BMI might be a better predictor of obesity-related health risk
318
in persons with MS with higher disability levels. In all groups, stronger associations were
319
observed between BMI and measures of body fat than between BMI and lean tissue, suggesting
320
that BMI is specific for adiposity. This is consistent with previous results in a sample of 95
321
patients with coronary artery disease that reported stronger associations between BMI and
322
%BF (ρ=.66) than BMI and lean mass (ρ=.41).9 Considering the relationship between BMI and
323
%BF by sex, we did not determine any differences in these associations. Previous research in
324
the general adult population has reported somewhat stronger associations between BMI and
325
%BF in women (n=7021, ρ=.87) compared to men (n=6580, ρ=.65), although the associations
326
were strong regardless of sex.2 Collectively, these findings support a potentially different
327
association between BMI and body composition in clinical populations, emphasizing the
328
importance of population-specific examinations of this relationship.
AC C
EP
TE D
M AN U
SC
313
329 330
The mechanisms underlying changes in body composition in persons with MS are likely related
331
to processes associated with the disease itself (e.g., chronic inflammation, glucocorticoid usage,
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 16
vitamin D deficiency)10,34,35 as well as physical activity behaviors,36–38 and accordingly, such
333
mechanisms may change with disease and disability progression over time. Indeed, individuals
334
with MS participate in substantially less physical activity than those without MS,36,37 and
335
persons with MS spend approximately 7.5 hours per day sitting.38 Inactivity and excessive
336
sitting likely contribute to shifting in body compartments resulting in the accumulation of
337
adipose tissue and the loss of bone and lean tissue mass. Similar shifts in body composition
338
have been noted with aging,39–41 and in other neurologic populations.42,43 Physical activity levels
339
and sedentary time are associated with disability status in persons with MS,36,38,44–46 and likely
340
contribute to additional changes in body composition with disease progression, although there
341
has yet to be a comprehensive examination of body composition across the disability spectrum
342
in persons with MS. The accumulation of additional adipose tissue and loss of bone and lean
343
mass with disability progression might explain the stronger association observed between BMI
344
and obesity in persons with MS with gait disability compared to those without gait disability in
345
this study.
347
349 350
Study Limitations
AC C
348
EP
346
TE D
M AN U
SC
RI PT
332
351
There were several strengths of this cross-sectional investigation including the large sample,
352
inclusion of a matched, non-MS reference sample, and the assessment of body composition
353
using gold standard measurement (i.e., DXA). This study was limited in that we performed a
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 17
secondary analysis of existing data from several investigations that were not originally designed
355
to examine the relationship between BMI and obesity, although body composition assessments
356
were conducted consistently across all studies. The assessment of body composition was not
357
conducted in a fasted state and this might have confounded our results. There is evidence to
358
suggest consuming a small meal after an overnight fast does not impact body composition
359
assessed by DXA or bioelectrical impedance in older adults,47 although this has not been
360
examined in MS samples. We further acknowledge that the validity of DXA in the evaluation of
361
adipose tissue mass has not been established in persons with MS. Although participants were
362
screened for risk factors for exercise participation, it is possible that participants suffered from
363
body fluid content altering or body fluid shifting diseases, and this could have impacted body
364
composition results. Based on our screening criteria, we might have further excluded potential
365
participants with high levels of body fat and this might have biased our findings. We further
366
acknowledge that we did not control for potential confounding effects of medication or
367
physical activity levels. Finally, disability level was determined by self-report rather than clinical
368
examination, although there is support for the PDDS as a valid patient-reported measure of
369
disability.25 The results of this study are generalizable to individuals with MS who are primarily
370
female, 22-64 years of age (i.e., age range of MS sample overall), have a relapsing disease
371
course, and ambulatory with mild-to-moderate disability.
373 374 375
SC
M AN U
TE D
EP
AC C
372
RI PT
354
CONCLUSIONS
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 18
We provide the first study to comprehensively examine the use of BMI as a marker of obesity in
377
persons with MS. Although a strong association exists between BMI and adiposity, the current
378
BMI threshold for defining obesity underestimates true adiposity. Underestimating true
379
adiposity might result in delayed intervention and treatment, leading to negative health
380
consequences. Researchers and clinicians should be aware of the limitations of BMI as a tool for
381
defining obesity in MS, and should be cautious in the interpretation of current BMI thresholds
382
in this population.
SC
RI PT
376
M AN U
383 384 385 386
390 391 392 393 394 395 396 397
EP
389
AC C
388
TE D
387
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 19
References
398
1.
Report of a WHO Committee. World Health Organ Tech Rep Ser 1995;854:1–452.
400 401
World Health Organization. Physical status: the use and interpretation of anthropometry.
2.
RI PT
399
Romero-Corral A, Somers VK, Sierra-Johnson J, Thomas RJ, Collazo-Clavell ML, Korinek J, et al. Accuracy of body mass index in diagnosing obesity in the adult general population. Int J
403
Obes 2008;32:959–66.
404
3.
SC
402
Gómez-Ambrosi J, Silva C, Galofré JC, Escalada J, Santos S, Millán D, et al. Body mass index classification misses subjects with increased cardiometabolic risk factors related to
406
elevated adiposity. Int J Obes 2012;36:286–94.
407
4.
M AN U
405
Okorodudu DO, Jumean MF, Montori VM, Romero-Corral A, Somers VK, Erwin PJ, et al. Diagnostic performance of body mass index to identify obesity as defined by body
409
adiposity: a systematic review and meta-analysis. Int J Obes 2010;34:791–9.
410
5.
Wellens RI, Roche AF, Khamis HJ, Jackson AS, Pollock ML, Siervogel RM. Relationships between the body mass index and body composition. Obes Res 1996;4:35–44.
411
6.
Blew RM, Sardinha LB, Milliken LA, Teixeira PJ, Going SB, Ferreira DL, et al. Assessing the
EP
412
TE D
408
validity of body mass index standards in early postmenopausal women. Obes Res
414
2002;10:799–808.
415
7.
AC C
413
Laughton GE, Buchholz AC, Martin Ginis KA, Goy RE, SHAPE SCI Research Group. Lowering
416
body mass index cutoffs better identifies obese persons with spinal cord injury. Spinal
417
Cord 2009;47:757–62.
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 20 418
8.
Alschuler KN, Gibbons LE, Rosenberg DE, Ehde DM, Verrall AM, Bamer AM, et al. Body mass index and waist circumference in persons aging with muscular dystrophy, multiple
420
sclerosis, post-polio syndrome, and spinal cord injury. Disabil Health J 2012;5:177–84.
421
9.
RI PT
419
Romero-Corral A, Somers VK, Sierra-Johnson J, Jensen MD, Thomas RJ, Squires RW, et al. Diagnostic performance of body mass index to detect obesity in patients with coronary
423
artery disease. Eur Heart J 2007;28:2087–93.
424
SC
422
10. Hedström AK, Olsson T, Alfredsson L. High body mass index before age 20 is associated with increased risk for multiple sclerosis in both men and women. Mult Scler
426
2012;18:1334–6.
428 429
11. Munger KL, Chitnis T, Ascherio A. Body size and risk of MS in two cohorts of US women. Neurology 2009;73:1543–50.
12. Khurana SR, Bamer AM, Turner AP, Wadhwani RV, Bowen JD, Leipertz SL, et al. The
TE D
427
M AN U
425
430
prevalence of overweight and obesity in veterans with multiple sclerosis. Am J Phys Med
431
Rehabil 2009;88:83–91.
434 435
EP
433
13. Pilutti LA, Dlugonski D, Pula JH, Motl RW. Weight status in persons with multiple sclerosis: implications for mobility outcomes. J Obes 2012;2012:868256. 14. Marrie R, Horwitz R, Cutter G, Tyry T, Campagnolo D, Vollmer T. High frequency of adverse
AC C
432
health behaviors in multiple sclerosis. Mult Scler 2009;15:105–13.
436
15. Slawta JN, Wilcox AR, McCubbin JA, Nalle DJ, Fox SD, Anderson G. Health behaviors, body
437
composition, and coronary heart disease risk in women with multiple sclerosis. Arch Phys
438
Med Rehabil 2003;84:1823–30.
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 21
440 441
16. Yarar-Fisher C, Chen Y, Jackson AB, Hunter GR. Body mass index underestimates adiposity in women with spinal cord injury. Obes 2013;21:1223–5. 17. Paul L, Coote S, Crosbie J, Dixon D, Hale L, Holloway E, et al. Core outcome measures for
RI PT
439
442
exercise studies in people with multiple sclerosis: recommendations from a
443
multidisciplinary consensus meeting. Mult Scler 2014;20:1641–50.
18. Pilutti LA, Dlugonski D, Sandroff BM, Klaren RE, Motl RW. Internet-delivered lifestyle
SC
444
physical activity intervention improves body composition in multiple sclerosis: preliminary
446
evidence from a randomized controlled trial. Arch Phys Med Rehabil 2014;95:1283–8.
447
19. Mojtahedi MC, Snook EM, Motl RW, Evans EM. Bone health in ambulatory individuals with
M AN U
445
448
multiple sclerosis: impact of physical activity, glucocorticoid use, and body composition. J
449
Rehabil Res Dev 2008;45:851–61.
451
20. Ward CL, Suh Y, Lane AD, Yan H, Ranadive SM, Fernhall B, et al. Body composition and
TE D
450
physical function in women with multiple sclerosis. J Rehabil Res Dev 2013;50:1139–47. 21. Snook EM, Mojtahedi MC, Evans EM, McAuley E, Motl RW. Physical activity and body
453
composition among ambulatory individuals with multiple sclerosis. Int J MS Care
454
2005;7:137–42.
456 457 458 459 460
22. Thomas S, Reading J, Shephard RJ. Revision of the Physical Activity Readiness
AC C
455
EP
452
Questionnaire (PAR-Q). Can J Sport Sci 1992;17:338–45. 23. Hohol MJ, Orav EJ, Weiner HL. Disease steps in multiple sclerosis: a simple approach to evaluate disease progression. Neurology 1995;45:251–5. 24. Hadjimichael O, Kerns RD, Rizzo MA, Cutter G, Vollmer T. Persistent pain and uncomfortable sensations in persons with multiple sclerosis. Pain 2007;127:35–41.
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 22 461
25. Learmonth YC, Motl RW, Sandroff BM, Pula JH, Cadavid D. Validation of patient determined disease steps (PDDS) scale scores in persons with multiple sclerosis. BMC
463
Neurol 2013;13:37.
464
RI PT
462
26. Ensari I, Motl RW, McAuley E, Mullen SP, Feinstein A. Patterns and predictors of naturally occurring change in depressive symptoms over a 30-month period in multiple sclerosis.
466
Mult Scler 2014;20:602–9.
468 469
27. Motl RW, Mullen S, McAuley E. Multi-group measurement invariance of the multiple sclerosis walking scale-12? Neurol Res 2012;34:149–52.
M AN U
467
SC
465
28. Peterson MD, Gordon PM, Hurvitz EA. Chronic disease risk among adults with cerebral
470
palsy: the role of premature sarcopoenia, obesity and sedentary behaviour. Obes Rev
471
2013;14:171–82.
473 474
29. Cohen J. Statistical power analysis for the behavioral sciences. 2nd ed. Lawrence Erlbaum
TE D
472
Associates: Hillsdale, NJ, 1988.
30. Buchholz AC, Bugaresti JM. A review of body mass index and waist circumference as markers of obesity and coronary heart disease risk in persons with chronic spinal cord
476
injury. Spinal Cord 2005;43:513–8. 31. Ravensbergen HRJC, Lear SA, Claydon VE. Waist circumference is the best index for
AC C
477
EP
475
478
obesity-related cardiovascular disease risk in individuals with spinal cord injury. J
479
Neurotrauma 2013;31:292–300.
480
32. Haslam DW, James WPT. Obesity. Lancet 2005;366:1197–209.
481
33. Marrie RA, Horwitz RI. Emerging effects of comorbidities on multiple sclerosis. Lancet
482
Neurol 2010;9:820–8.
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 23 483
34. Formica CA, Cosman F, Nieves J, Herbert J, Lindsay R. Reduced bone mass and fat-free mass in women with multiple sclerosis: effects of ambulatory status and glucocorticoid
485
use. Calcif Tissue Int 1997;61:129–33.
RI PT
484
35. Ozgocmen S, Bulut S, Ilhan N, Gulkesen A, Ardicoglu O, Ozkan Y. Vitamin D deficiency and
487
reduced bone mineral density in multiple sclerosis: effect of ambulatory status and
488
functional capacity. J Bone Miner Metab 2005;23:309–13.
490 491 492 493
36. Motl RW, McAuley E, Sandroff BM, Hubbard EA. Descriptive epidemiology of physical activity rates in multiple sclerosis. Acta Neurol Scand 2015;131:422–5.
M AN U
489
SC
486
37. Motl RW, McAuley E, Snook EM. Physical activity and multiple sclerosis: a meta-analysis. Mult Scler 2005;11:459–63.
38. Hubbard EA, Motl RW, Manns PJ. The descriptive epidemiology of daily sitting time as a sedentary behavior in multiple sclerosis. Disabil Health J 2015;
495
doi:10.1016/j.dhjo.2015.06.003.
TE D
494
39. Sowers M, Zheng H, Tomey K, Karvonen-Gutierrez C, Jannausch M, Li X, et al. Changes in
497
body composition in women over six years at midlife: Ovarian and chronological aging. J
498
Clin Endocrinol Metab 2007;92:895–901.
500
40. Guo SS, Zeller C, Chumlea WC, Siervogel RM. Aging, body composition, and lifestyle: the
AC C
499
EP
496
Fels Longitudinal Study. Am J Clin Nutr 1999;70:405–11.
501
41. Fantin F, Francesco VD, Fontana G, Zivelonghi A, Bissoli L, Zoico E, et al. Longitudinal body
502
composition changes in old men and women: Interrelationships with worsening disability.
503
J Gerontol A Biol Sci Med Sci 2007;62:1375–81.
504
42. Kocina DP. Body composition of spinal cord injured adults. Sports Med 1997;23:48–60.
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 24
506 507 508 509
43. Jørgensen L, Jacobsen BK. Changes in muscle mass, fat mass, and bone mineral content in the legs after stroke: a 1 year prospective study. Bone 2001;28:655–9. 44. Motl RW, Goldman M. Physical inactivity, neurological disability, and cardiorespiratory
RI PT
505
fitness in multiple sclerosis. Acta Neurol Scand 2011;123:98–104.
45. Dlugonski D, Pilutti LA, Sandroff BM, Suh Y, Balantrapu S, Motl RW. Steps per day among persons with multiple sclerosis: variation by demographic, clinical, and device
511
characteristics. Arch Phys Med Rehabil 2013;94:1534–9.
46. Hubbard EA, Motl RW. Sedentary behavior is associated with disability status and walking
M AN U
512
SC
510
513
performance, but not cognitive function, in multiple sclerosis. Appl Physiol Nutr Metab
514
2015;40:203–6.
518 519 520 521 522 523 524 525 526
TE D
517
the body composition evaluation of elderly persons. J Nutr Health Aging 2009;13:183–6.
EP
516
47. Vilaça KHC, Ferriolli E, Lima NKC, Paula FJA, Moriguti JC. Effect of fluid and food intake on
AC C
515
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 25 527
Suppliers
529
a. Detecto, 203 E Daugherty St., Webb City, MO 64870.
530
b. Hologic, Inc., 35 Crosby Dr., Bedford, MA 01730.
531
c. SPSS Inc., 233 S Wacker Dr, 11th Fl, Chicago, IL 60606.
SC
532 533
M AN U
534 535 536 537
542 543 544 545 546 547 548
EP
541
AC C
540
TE D
538 539
RI PT
528
ACCEPTED MANUSCRIPT Body mass index in multiple sclerosis 26 549
Figure Legend
550
Fig 1 Receiver-operating characteristic (ROC) curves for BMI to detect sex-specific %BF-define
552
obesity for non-MS controls, participants with MS overall, and by level of disability. (a) non-MS
553
controls: AUC=.833, specificity=94%, sensitivity=37%; (b) MS sample overall: AUC=.846,
554
specificity=93%, sensitivity=44%; (c) PDDS<3.0: AUC=.815, specificity=93%, sensitivity=44%; and
555
(d) PDDS≥3.0: AUC=.913, specificity=100%, sensitivity=44%.
AC C
EP
TE D
M AN U
SC
RI PT
551
ACCEPTED MANUSCRIPT
Table 1 Demographic and clinical characteristics of non-MS controls, and participants with MS overall and by level of disability. MS
P value
Overall
PDSS <3.0
PDDS ≥3.0
Control vs
PDDS
(n=53)
(n=235)
(n=124)
(n=103)
MS Overall
groups
49.2 (11.1)
48.1 (9.3)
45.7 (9.8)
50.8 (8.0)
.45
<.001
43/10
181/54
96/28
78/25
.52
.76
Height, cm
166.6 (9.4)
167.8 (8.9)
168.1 (9.1)
167.2 (8.8)
.40
.49
Weight, kg
73.0 (17.3)
77.8 (19.4)
78.5 (18.6)
76.3 (19.3)
.10
.38
PDDS, mdn (IQR)*
N/A
2.0 (3.0)
1.0 (1.0)
4.0 (2.0)
N/A
<.001
Disease duration, years*
N/A
10.8 (7.9)
9.2 (7.2)
12.9 (8.5)
N/A
<.001
Disease course,
N/A
194/41
117/7
70/33
N/A
<.001
Sex, female/male
TE D
relapsing/progressive*
SC
Age, years*
M AN U
Characteristic
RI PT
Control
BMI=body mass index. IQR=interquartile range. mdn=median. Values are mean (SD), unless otherwise
AC C
EP
noted. *Statistically significant difference between MS disability groups (i.e., PDDS <3.0 vs. PDDS ≥3.0).
ACCEPTED MANUSCRIPT
Table 2 Morphological characteristics of non-MS controls, and participants with MS overall and by level of disability.
Control (n=53)
Overall (n=235)
PDSS <3.0 (n=124)
PDDS ≥3.0 (n=103)
26.4 (6.5)
27.7 (6.9)
27.9 (7.1)
27.2 (6.3)
Under-normal, n (%)
27 (50.9%)
105 (44.7%)
SC
Characteristic
RI PT
MS
46 (44.7%)
Overweight, n (%)
15 (28.3%)
61 (26.0%)
32 (27.2%)
28 (27.2%)
Obese, n (%)
11 (20.8%)
69 (29.0%)
37 (29.8%)
29 (28.2%)
31.9 (10.8)
35.0 (8.8)
34.7 (8.6)
35.3 (9.2)
27 (50.9%)
139 (59.1%)
71 (57.3%)
65 (63.1%)
28,260.1 (12,943.0)
28,359.4 (12,289.5)
28,004.2 (12,913.2)
47,175.4 (9,534.7)
48,068.2 (9,194.1)
46,175.8 (9,489.8)
2,299.6 (350.1)
2,358.7 (344.7)
2,234.5 (344.4)
1.122 (0.092)
1.137 (0.095)
1.106 (0.087)
BMI, kg/m2
Obese, n (%)
24,6334.0 (12,607.6)
Whole body LST mass, g
48,019.6 (10,310.5)
Whole body BMD, g/cm2
2,318.9 (438.5)
AC C
Whole body BMC, g*
EP
Whole body fat mass, g
TE D
Whole body fat, (%)
55 (44.4%)
M AN U
BMI classification
1.124 (0.121)
BMC= bone mineral content. BMD= bone mineral density. BMI=body mass index. LST= lean soft tissue. Values are mean (SD), unless otherwise noted. *Statistically significant difference between MS disability groups (i.e., PDDS <3.0 vs. PDDS ≥3.0).
ACCEPTED MANUSCRIPT
Table 3 Bivariate Pearson and Spearman correlation coefficients between BMI and DXA outcomes in non-MS controls, and participants with MS
RI PT
overall and by level of disability. MS
PDDS ≥3.0
(n=235)
(n=124)
(n=103)
Control (n=53)
Percent body fat^
.61* (.41-.76)
.65* (.46-.78) Whole body fat mass, g#
.75* (.65-.82)
.83* (.79-.87)
.76* (.67-.83)
.92* (.88-.95)
.80* (.75-.84)
.72* (.62-.80)
.92* (.88-.95)
.41* (.16-.61)
.46* (.35-.56)
.40* (.24-.54)
.52* (.36-.65)
.39* (.13-.60)
.48* (.38-.57)
.44* (.29-.57)
.56* (.41-.68)
.34* (.08-.56)
.23* (.11-.35)
.10 (-.08-.27)
.39* (.21-.54)
.27* (.15-.39)
.13 (-.05-.30)
.42* (.25-.57)
.37* (.11-.58)
.23* (.11-.35)
.25* (.08-.41)
.22* (.03-.40)
.46* (.22-.65)
.21* (.08-.33)
.17 (-.01-.34)
.24* (.05-.41)
TE D
EP AC C
.42* (.17-.62)
Whole body bone mineral density, g/cm2
.75* (.65-.82)
.61* (.49-.71)
.80* (.68-.88)
Whole body bone mineral content, g#
.59* (.46-.69)
.66* (.58-.73)
.89* (.82-.94)
Whole body lean soft tissue mass, g
.65* (.57-.72)
SC
PDSS <3.0
M AN U
Characteristic
Overall
ACCEPTED MANUSCRIPT
2 Spearman correlation coefficients are presented below Pearson. 95% confidence intervals are presented in brackets. *Statistically significant
RI PT
correlation coefficient. #Statistically significant difference in Pearson and Spearman correlation coefficients between MS disability groups (i.e., PDDS <3.0 vs. PDDS ≥3.0). ^Statistically significant difference in Pearson correlation coefficients between MS disability groups (i.e., PDDS <3.0 vs.
AC C
EP
TE D
M AN U
SC
PDDS ≥3.0).
ACCEPTED MANUSCRIPT
Table 4 Predicted %BF associated with standard BMI thresholds based on regression equations in non-MS controls, and participants with MS
RI PT
overall and by level of disability.
Overall
PDSS <3.0
PDDS ≥3.0
(kg/m2)
(n=53)
(n=235)
(n=124)
(n=103)
18.5
23.9
27.4
28.0
25.9
25
30.2
32.8
32.7
33.0
30
35.6
36.9
36.3
38.4
M AN U
Control
AC C
EP
TE D
BMI threshold
SC
MS
ACCEPTED MANUSCRIPT
a
M AN U
SC
RI PT
b
d
AC C
EP
TE D
c