research letter
Diabetes, Obesity and Metabolism 17: 91–93, 2015. © 2014 John Wiley & Sons Ltd
There is concern that intentional weight loss may generate excessive loss of fat-free mass (FFM). Idealists target minimal loss of FFM, while others consider that FFM loss of up to 25% of weight loss is acceptable. In a cross-sectional study of 275 weight-stable, overweight or obese adults, we used whole-body dual-energy X-ray absorptiometry to measure FFM. A range of models was used to estimate the expected ΔFFM/Δweight ratio required to attain the body composition of a weight-stable individual at a lower body mass index (BMI). Higher BMI was associated linearly with higher FFM in men and women. Proportional ΔFFM/Δweight was influenced by sex, BMI and age. Direct scatter plot analysis, quadratic curve fit modelling and linear FFM–BMI modelling provided similar estimates for each model of ΔFFM/Δweight ratio, with 40% for men and 33% for women. These results show that the 25% rule is inappropriate and our estimates are higher than those generally reported after intentional weight loss indicating favourable preservation of FFM. Keywords: body composition, treatment guidelines, observational study, fat Date submitted 25 July 2014; date of first decision 18 August 2014; date of final acceptance 2 September 2014
Introduction The importance of maintaining fat-free mass (FFM), much of which is muscle mass, with any substantial intentional weight loss is reflected in the need to maintain resting energy expenditure, strength and function and to prevent sarcopenia and frailty. In the past, a change in FFM to change in body weight ratio (ΔFFM/Δweight) of 25 kg/m2 ), weight-stable for 6 months (±2 kg), and Correspondence to : John Dixon, MBBS, PhD, FRACGP, FRCP Edin, Baker IDI Heart and Diabetes Institute, 75 Commercial Road, PO Box 6492, Melbourne, Victoria 3004, Australia. E-mail: [email protected]
not participating in a weight loss programme. All studies were approved by the Alfred Hospital, Human Ethical Review Committee. Participants underwent a baseline whole-body DEXA scan (GE-LUNAR Prodigy Advance PA_130510; GE Medical Systems, Lunar, Madison, WI, USA). The correlation between actual weight and DEXA weight was excellent at 99.1%. DEXA-derived weight was used in the analysis. We compared our results with those derived from the Pennington Fat-Free Mass calculator, based on USA data obtained from several cohorts, including the National Health and Nutrition Examination Survey (NHANES) cohort, and incorporating race, weight, height and gender [4,5].
Statistical Analyses Scatter plots were used to examine the relationships between the participants’ BMI and FM and FFM. Regression analysis was used to assess individual characteristics influencing FFM and percentage of FFM. Locally weighted scatter plot smoothing (LOESS) was used to examine the non-linear complex relationships. Co-ordinates for the LOESS line plots for men and women were taken at each BMI level and direct calculations of the expected ΔFFM/Δweight percentages were made based on average heights for men and women. Linear, inverse, cubic and quadratic equations were used to assess best-fit trend curves. In addition, linear relationships of FM and FFM with BMI were used to calculate expected changes in percentage of FFM with modelled weight loss.
Results We studied 275 adults (138 women and 137 men) of European ethnicity, whose mean ± standard deviation (s.d.)
RESEARCH LETTER
Fat-free mass loss generated with weight loss in overweight and obese adults: what may we expect?
research letter
DIABETES, OBESITY AND METABOLISM
age was 45.5 ± 11.6 years and whose mean ± s.d. BMI was 36.4 ± 7 kg/m2 (see Table S1, Supporting Information for full details, including body composition). The DEXA-derived FFM (kg) was influenced independently by body weight, gender, age and height in descending order of importance, and together these factors explained 90% of FFM (kg) variance in this cohort (Figure S1, Supporting Information). As weight and height could be replaced by BMI with minimal loss of explained variance (90–89%), our modelling used BMI for the expected ΔFFM/Δweight percentage. There was a positive linear association between BMI and both total body fat and FFM in men and women (Figures S2A and B). Total FM was surprisingly similar for both men and women for a given BMI (Figure S2A), but the total FFM was very different (Figure S2B).
Estimates of Percentage of Fat-Free Mass Loss Using Locally Weighted Scatter Plot Smoothing Lines For the overall group, gender, age and BMI explained 77% of variance in the percentage of FFM. The relationship between BMI and the percentage of body weight that is fat (%FM) and FFM (%FFM) is not linear, as indicated by the 50% LOESS lines (Figure 1A and B). The BMI and %FM relationship was curvilinear. The estimated proportion of FFM loss varied with gender, initial BMI and the extent of weight loss. For men, the proportion of FFM loss varied from 35.0 to 45.5% (mean 40.1%). Values for women ranged from 28 to 41.2% (mean 35.9%). Table 1 shows examples of the estimated decrease in %FFM with a 10% decrease in total body weight for varying baseline BMI values, using several methods of estimation as described below.
Estimates of Percentage of Fat-Free Mass Loss Data transformation using 1/BMI improved curve fit for men, but did not adequately account for the relationship in women with BMI >35 kg/m2 . The 1/BMI analysis generated a constant proportion of FFM loss of 37.1% for men and 29.5% for women (Table 1). Quadratic curve estimation provided the best fit with the data and closely resembled the LOESS fit line for both men and women. Analysis of a 10% weight loss throughout a large BMI range showed an expected average FFM loss of 41.3 and 33.2% of total weight loss for men and women, respectively (Table 1). An alternative method of estimation was derived from the linear relationships of FM and FFM with BMI for both men and women (Figure S2). Using the regression line equations, a constant expected loss of FFM was calculated to be 40.9% of total weight loss for men and 33.9% for women (Table 1).
Comparison of Percentage of Fat-Free Mass Loss Using US Population Calculator The FFM for each participant was estimated using the FFM calculator developed by Thomas et al. [4] using US data for those of white race and incorporating weight, height and gender. Our individual DEXA-derived FFM measures exceeded the US calculator estimates slightly, with a mean difference of 0.6 ± 5.3 kg, which was not statistically significant.
92 Dixon et al.
Figure 1. The percentage of total body weight that is (A) fat mass and (B) fat-free mass for adult men and women in relation to their body mass index (BMI; n = 275).
Discussion Data from the actual 50% LOESS line curves, the regression generated quadratic curve and linear increases in FM and FFM with increasing BMI all provided similar estimates for the expected ΔFFM/Δweight percentage, with ∼40 and 33% of the weight loss expected to be FFM loss for men and women, respectively. A limitation of the present study is the fact that the participants were volunteers eligible for various research studies and may therefore include more engaged and able individuals who were generally without severe or complex comorbidities. In addition, all the participants were white and of European ethnicity. The study confirms the clear relationship of FM and FFM with increasing BMI and is consistent with the cross-sectional
Volume 17 No. 1 January 2015
research letter
DIABETES, OBESITY AND METABOLISM
Table 1. Estimated ratios of decrease in fat-free mass to total decrease in body weight (%) for various baseline body mass indices required to achieve the usual body composition at a 10% lower body weight for men and women. Decrease in FFM to total decrease in body weight ratio, % Baseline BMI Men Actual Inverse Quadratic FFM linear Women Actual Inverse Quadratic FFM linear
BMI 50 kg/m2
BMI 45 kg/m2
BMI 40 kg/m2
BMI 35 kg/m2
BMI 30 kg/m2
R2
Mean
38.7 37.1 44.8 40.9
38.7 37.1 40.7 40.9
45.3 37.1 38.9 40.9
38.1 37.1 39.3 40.9
42.2 37.1 42.4 40.9
— 0.45 0.46 —
40.6 37.1 41.2 40.9
37.7 29.5 42.7 33.9
39.4 29.5 35.2 33.9
38.7 29.5 30.1 33.9
28.4 29.5 28.1 33.9
35.7 29.5 30 33.9
— 0.33 0.36 —
36.0 29.5 33.2 33.9
BMI, body mass index; FFM, fat-free mass. R2 values derived from curve using all data for men and women respectively. Actual percentages for FFM were calculated using the points on the locally weighted scatter plot smoothing lines. Inverse and quadratic estimates were based on best-fit lines. FFM linear estimates were calculated using the linear approximation of the decrease in FFM shown in Figure S2A. Average heights of 1.78 m for men and 1.63 m for women, respectively, were used in all calculations.
and longitudinal overfeeding experiments by Forbes et al. [6,7], which indicated that as much as 40% of additional weight loss or weight gain was FFM. The present findings are also broadly consistent with the new US population DEXA-derived formula for calculating FFM [4]. There is now clear evidence that the earlier 25% ΔFFM/Δweight approach for estimating FFM should be replaced by more realistic individualized measures. With most methods of weight loss, actual changes generate greater preservation of FFM than expected for the change in BMI [3]. This is reassuring for the problem of weight cycling in obese individuals and should not discourage future attempts to lose weight and improve health [8,9]. In conclusion, there is a robust relationship between FFM and FM with increasing BMI. With intentional weight loss, considerable loss of FFM is expected if an individual is to attain the expected body composition at the lower BMI. As a guide, 35–40% and 30–35% are the expected FFM to total weight loss ratios for men and women, respectively, and anything less can be seen as a bonus. J. B. Dixon1,2 , E. A. Lambert1,3,4 , M. Grima1,3 , T. Rice1,3 , G. W. Lambert3,5 & N. E. Straznicky1,3 1 Clinical Obesity Research, Baker IDI Heart & Diabetes Institute, Melbourne, Australia 2 Primary Care Research Unit, Monash University, Melbourne, Australia 3 Human Neurotransmitters Laboratories, Baker IDI Heart & Diabetes Institute, Melbourne, Australia 4 Department of Physiology, Monash University, Melbourne, Australia 5 Department of Medicine, Monash University, Melbourne, Australia
Conflict of Interest No conflict of interest declared. J. B. D. was involved in all aspects of the study. E. A. L. contributed to the study design and conduct, and collection of data. M. G. contributed to the study conduct and data collection. T. R. contributed to
Volume 17 No. 1 January 2015
the study design and conduct, data collection and writing. G. W. L. contributed to the study design and conduct, and data collection and analysis. N. E. S. contributed to the study design and conduct, and data collection.
Supporting Information Additional Supporting Information may be found in the online version of this article: Table S1. Characteristics of all participants (n = 275) attending for baseline study examinations. Figure S1. The relationship between actual DEXA-measured FFM and modelled FFM for each individual based on weight, gender, age and height. These factors explained 90% of variance in FFM. Figure S2. Total fat free mass (kg) for adult men and women in relation to their body mass index (BMI) (n = 275).
References 1. Webster JD, Hesp R, Garrow JS. The composition of excess weight in obese women estimated by body density, total body water and total body potassium. Hum Nutr Clin Nutr 1984; 38: 299–306. 2. Heymsfield SB, Gonzalez MC, Shen W, Redman L, Thomas D. Weight loss composition is one-fourth fat-free mass: a critical review and critique of this widely cited rule. Obes Rev 2014; 15: 310–321. 3. Chaston TB, Dixon JB, O’Brien PE. Changes in fat-free mass during significant weight loss: a systematic review. Int J Obes (Lond) 2007; 31: 743–750. 4. Thomas D, Das SK, Levine JA et al. New fat free mass – fat mass model for use in physiological energy balance equations. Nutr Metab (Lond) 2010; 7: 39. 5. PBRC. Change lean mass calculator. 2010. Available from URL: http://pbrc. edu/research-and-faculty/calculators/fat-free-mass/. Accessed 22 June 2014. 6. Forbes GB, Welle SL. Lean body mass in obesity. Int J Obes 1983; 7: 99–107. 7. Forbes GB, Brown MR, Welle SL, Lipinski BA. Deliberate overfeeding in women and men: energy cost and composition of the weight gain. Br J Nutr 1986; 56: 1–9. 8. Mason C, Foster-Schubert KE, Imayama I et al. History of weight cycling does not impede future weight loss or metabolic improvements in postmenopausal women. Metabolism 2013; 62: 127–136. 9. Casazza K, Fontaine KR, Astrup A et al. Myths, presumptions, and facts about obesity. N Engl J Med 2013; 368: 446–454.
doi:10.1111/dom.12389 93
Diabetes, Obesity and Metabolism 17: 91–93, 2015. © 2014 John Wiley & Sons Ltd
There is concern that intentional weight loss may generate excessive loss of fat-free mass (FFM). Idealists target minimal loss of FFM, while others consider that FFM loss of up to 25% of weight loss is acceptable. In a cross-sectional study of 275 weight-stable, overweight or obese adults, we used whole-body dual-energy X-ray absorptiometry to measure FFM. A range of models was used to estimate the expected ΔFFM/Δweight ratio required to attain the body composition of a weight-stable individual at a lower body mass index (BMI). Higher BMI was associated linearly with higher FFM in men and women. Proportional ΔFFM/Δweight was influenced by sex, BMI and age. Direct scatter plot analysis, quadratic curve fit modelling and linear FFM–BMI modelling provided similar estimates for each model of ΔFFM/Δweight ratio, with 40% for men and 33% for women. These results show that the 25% rule is inappropriate and our estimates are higher than those generally reported after intentional weight loss indicating favourable preservation of FFM. Keywords: body composition, treatment guidelines, observational study, fat Date submitted 25 July 2014; date of first decision 18 August 2014; date of final acceptance 2 September 2014
Introduction The importance of maintaining fat-free mass (FFM), much of which is muscle mass, with any substantial intentional weight loss is reflected in the need to maintain resting energy expenditure, strength and function and to prevent sarcopenia and frailty. In the past, a change in FFM to change in body weight ratio (ΔFFM/Δweight) of 25 kg/m2 ), weight-stable for 6 months (±2 kg), and Correspondence to : John Dixon, MBBS, PhD, FRACGP, FRCP Edin, Baker IDI Heart and Diabetes Institute, 75 Commercial Road, PO Box 6492, Melbourne, Victoria 3004, Australia. E-mail: [email protected]
not participating in a weight loss programme. All studies were approved by the Alfred Hospital, Human Ethical Review Committee. Participants underwent a baseline whole-body DEXA scan (GE-LUNAR Prodigy Advance PA_130510; GE Medical Systems, Lunar, Madison, WI, USA). The correlation between actual weight and DEXA weight was excellent at 99.1%. DEXA-derived weight was used in the analysis. We compared our results with those derived from the Pennington Fat-Free Mass calculator, based on USA data obtained from several cohorts, including the National Health and Nutrition Examination Survey (NHANES) cohort, and incorporating race, weight, height and gender [4,5].
Statistical Analyses Scatter plots were used to examine the relationships between the participants’ BMI and FM and FFM. Regression analysis was used to assess individual characteristics influencing FFM and percentage of FFM. Locally weighted scatter plot smoothing (LOESS) was used to examine the non-linear complex relationships. Co-ordinates for the LOESS line plots for men and women were taken at each BMI level and direct calculations of the expected ΔFFM/Δweight percentages were made based on average heights for men and women. Linear, inverse, cubic and quadratic equations were used to assess best-fit trend curves. In addition, linear relationships of FM and FFM with BMI were used to calculate expected changes in percentage of FFM with modelled weight loss.
Results We studied 275 adults (138 women and 137 men) of European ethnicity, whose mean ± standard deviation (s.d.)
RESEARCH LETTER
Fat-free mass loss generated with weight loss in overweight and obese adults: what may we expect?
research letter
DIABETES, OBESITY AND METABOLISM
age was 45.5 ± 11.6 years and whose mean ± s.d. BMI was 36.4 ± 7 kg/m2 (see Table S1, Supporting Information for full details, including body composition). The DEXA-derived FFM (kg) was influenced independently by body weight, gender, age and height in descending order of importance, and together these factors explained 90% of FFM (kg) variance in this cohort (Figure S1, Supporting Information). As weight and height could be replaced by BMI with minimal loss of explained variance (90–89%), our modelling used BMI for the expected ΔFFM/Δweight percentage. There was a positive linear association between BMI and both total body fat and FFM in men and women (Figures S2A and B). Total FM was surprisingly similar for both men and women for a given BMI (Figure S2A), but the total FFM was very different (Figure S2B).
Estimates of Percentage of Fat-Free Mass Loss Using Locally Weighted Scatter Plot Smoothing Lines For the overall group, gender, age and BMI explained 77% of variance in the percentage of FFM. The relationship between BMI and the percentage of body weight that is fat (%FM) and FFM (%FFM) is not linear, as indicated by the 50% LOESS lines (Figure 1A and B). The BMI and %FM relationship was curvilinear. The estimated proportion of FFM loss varied with gender, initial BMI and the extent of weight loss. For men, the proportion of FFM loss varied from 35.0 to 45.5% (mean 40.1%). Values for women ranged from 28 to 41.2% (mean 35.9%). Table 1 shows examples of the estimated decrease in %FFM with a 10% decrease in total body weight for varying baseline BMI values, using several methods of estimation as described below.
Estimates of Percentage of Fat-Free Mass Loss Data transformation using 1/BMI improved curve fit for men, but did not adequately account for the relationship in women with BMI >35 kg/m2 . The 1/BMI analysis generated a constant proportion of FFM loss of 37.1% for men and 29.5% for women (Table 1). Quadratic curve estimation provided the best fit with the data and closely resembled the LOESS fit line for both men and women. Analysis of a 10% weight loss throughout a large BMI range showed an expected average FFM loss of 41.3 and 33.2% of total weight loss for men and women, respectively (Table 1). An alternative method of estimation was derived from the linear relationships of FM and FFM with BMI for both men and women (Figure S2). Using the regression line equations, a constant expected loss of FFM was calculated to be 40.9% of total weight loss for men and 33.9% for women (Table 1).
Comparison of Percentage of Fat-Free Mass Loss Using US Population Calculator The FFM for each participant was estimated using the FFM calculator developed by Thomas et al. [4] using US data for those of white race and incorporating weight, height and gender. Our individual DEXA-derived FFM measures exceeded the US calculator estimates slightly, with a mean difference of 0.6 ± 5.3 kg, which was not statistically significant.
92 Dixon et al.
Figure 1. The percentage of total body weight that is (A) fat mass and (B) fat-free mass for adult men and women in relation to their body mass index (BMI; n = 275).
Discussion Data from the actual 50% LOESS line curves, the regression generated quadratic curve and linear increases in FM and FFM with increasing BMI all provided similar estimates for the expected ΔFFM/Δweight percentage, with ∼40 and 33% of the weight loss expected to be FFM loss for men and women, respectively. A limitation of the present study is the fact that the participants were volunteers eligible for various research studies and may therefore include more engaged and able individuals who were generally without severe or complex comorbidities. In addition, all the participants were white and of European ethnicity. The study confirms the clear relationship of FM and FFM with increasing BMI and is consistent with the cross-sectional
Volume 17 No. 1 January 2015
research letter
DIABETES, OBESITY AND METABOLISM
Table 1. Estimated ratios of decrease in fat-free mass to total decrease in body weight (%) for various baseline body mass indices required to achieve the usual body composition at a 10% lower body weight for men and women. Decrease in FFM to total decrease in body weight ratio, % Baseline BMI Men Actual Inverse Quadratic FFM linear Women Actual Inverse Quadratic FFM linear
BMI 50 kg/m2
BMI 45 kg/m2
BMI 40 kg/m2
BMI 35 kg/m2
BMI 30 kg/m2
R2
Mean
38.7 37.1 44.8 40.9
38.7 37.1 40.7 40.9
45.3 37.1 38.9 40.9
38.1 37.1 39.3 40.9
42.2 37.1 42.4 40.9
— 0.45 0.46 —
40.6 37.1 41.2 40.9
37.7 29.5 42.7 33.9
39.4 29.5 35.2 33.9
38.7 29.5 30.1 33.9
28.4 29.5 28.1 33.9
35.7 29.5 30 33.9
— 0.33 0.36 —
36.0 29.5 33.2 33.9
BMI, body mass index; FFM, fat-free mass. R2 values derived from curve using all data for men and women respectively. Actual percentages for FFM were calculated using the points on the locally weighted scatter plot smoothing lines. Inverse and quadratic estimates were based on best-fit lines. FFM linear estimates were calculated using the linear approximation of the decrease in FFM shown in Figure S2A. Average heights of 1.78 m for men and 1.63 m for women, respectively, were used in all calculations.
and longitudinal overfeeding experiments by Forbes et al. [6,7], which indicated that as much as 40% of additional weight loss or weight gain was FFM. The present findings are also broadly consistent with the new US population DEXA-derived formula for calculating FFM [4]. There is now clear evidence that the earlier 25% ΔFFM/Δweight approach for estimating FFM should be replaced by more realistic individualized measures. With most methods of weight loss, actual changes generate greater preservation of FFM than expected for the change in BMI [3]. This is reassuring for the problem of weight cycling in obese individuals and should not discourage future attempts to lose weight and improve health [8,9]. In conclusion, there is a robust relationship between FFM and FM with increasing BMI. With intentional weight loss, considerable loss of FFM is expected if an individual is to attain the expected body composition at the lower BMI. As a guide, 35–40% and 30–35% are the expected FFM to total weight loss ratios for men and women, respectively, and anything less can be seen as a bonus. J. B. Dixon1,2 , E. A. Lambert1,3,4 , M. Grima1,3 , T. Rice1,3 , G. W. Lambert3,5 & N. E. Straznicky1,3 1 Clinical Obesity Research, Baker IDI Heart & Diabetes Institute, Melbourne, Australia 2 Primary Care Research Unit, Monash University, Melbourne, Australia 3 Human Neurotransmitters Laboratories, Baker IDI Heart & Diabetes Institute, Melbourne, Australia 4 Department of Physiology, Monash University, Melbourne, Australia 5 Department of Medicine, Monash University, Melbourne, Australia
Conflict of Interest No conflict of interest declared. J. B. D. was involved in all aspects of the study. E. A. L. contributed to the study design and conduct, and collection of data. M. G. contributed to the study conduct and data collection. T. R. contributed to
Volume 17 No. 1 January 2015
the study design and conduct, data collection and writing. G. W. L. contributed to the study design and conduct, and data collection and analysis. N. E. S. contributed to the study design and conduct, and data collection.
Supporting Information Additional Supporting Information may be found in the online version of this article: Table S1. Characteristics of all participants (n = 275) attending for baseline study examinations. Figure S1. The relationship between actual DEXA-measured FFM and modelled FFM for each individual based on weight, gender, age and height. These factors explained 90% of variance in FFM. Figure S2. Total fat free mass (kg) for adult men and women in relation to their body mass index (BMI) (n = 275).
References 1. Webster JD, Hesp R, Garrow JS. The composition of excess weight in obese women estimated by body density, total body water and total body potassium. Hum Nutr Clin Nutr 1984; 38: 299–306. 2. Heymsfield SB, Gonzalez MC, Shen W, Redman L, Thomas D. Weight loss composition is one-fourth fat-free mass: a critical review and critique of this widely cited rule. Obes Rev 2014; 15: 310–321. 3. Chaston TB, Dixon JB, O’Brien PE. Changes in fat-free mass during significant weight loss: a systematic review. Int J Obes (Lond) 2007; 31: 743–750. 4. Thomas D, Das SK, Levine JA et al. New fat free mass – fat mass model for use in physiological energy balance equations. Nutr Metab (Lond) 2010; 7: 39. 5. PBRC. Change lean mass calculator. 2010. Available from URL: http://pbrc. edu/research-and-faculty/calculators/fat-free-mass/. Accessed 22 June 2014. 6. Forbes GB, Welle SL. Lean body mass in obesity. Int J Obes 1983; 7: 99–107. 7. Forbes GB, Brown MR, Welle SL, Lipinski BA. Deliberate overfeeding in women and men: energy cost and composition of the weight gain. Br J Nutr 1986; 56: 1–9. 8. Mason C, Foster-Schubert KE, Imayama I et al. History of weight cycling does not impede future weight loss or metabolic improvements in postmenopausal women. Metabolism 2013; 62: 127–136. 9. Casazza K, Fontaine KR, Astrup A et al. Myths, presumptions, and facts about obesity. N Engl J Med 2013; 368: 446–454.
doi:10.1111/dom.12389 93
