Article Text

Pitfalls in the assessment of body composition in survivors of acute lymphoblastic leukaemia
1. J T Warner1,
2. W D Evans2,
3. D K H Webb3,
4. J W Gregory4
1. 1Department of Paediatrics, John Radcliffe Hospital, Oxford, UK
2. 2Department of Medical Physics, University Hospital of Wales Healthcare NHS Trust, Heath Park, Cardiff, UK
3. 3Department of Child Health, Llandough Hospital NHS Trust, Penlan Road, Cardiff, UK
4. 4Department of Child Health, University of Wales College of Medicine, Heath Park, Cardiff, UK
1. Correspondence to:
Dr J T Warner
Department of Paediatrics, John Radcliffe Hospital, Oxford OX3 9DU, UK; justin.warnerorh.nhs.uk

## Abstract

Background: Body fat mass (FM) and fat free mass (FFM) in childhood are often estimated by conversion of a measured variable into compartmental body composition using constants or regression equations that have been previously derived in healthy individuals. Application of such constants or equations to children with disease states may lead to inappropriate conclusions since the “normal” relationships may become altered.

Aims and Methods: To test this hypothesis by taking measurements of body composition using dual energy x ray absorptiometry (DEXA) as a “gold standard” method and calculating hydration and body potassium constants using isotopic water dilution and whole body potassium counting. Measurements of bioelectrical impedance (BIA) by two different analysers (RJL and Holtain) were also performed to allow comparison with body water measurements.

Results: Measurements were performed in 35 children treated for acute lymphoblastic leukaemia (ALL) and compared to those in 21 children treated for a variety of other malignancies and 32 healthy sibling controls. The mean hydration and potassium content of FFM was significantly reduced in the ALL group compared to both other malignancies and controls. Application of equations derived from controls for the measurement of FFM derived from bioelectrical impedance led to an underestimation of 1.15 kg when compared to that derived from DEXA in children treated for ALL but not in other malignancies. For all groups combined, BIA was significantly different in the two analysers.

Conclusion: Care needs to be taken in the application of equations derived from the normal population to body composition measurement in children treated for ALL.

• body composition
• dual energy X-ray absorptiometry
• whole body potassium
• hydration constant
• ALL, acute lymphoblastic leukaemia
• AML, acute myeloid leukaemia
• BIA, bioelectrical impedance
• BMI, body mass index
• CNS, central nervous system
• DEXA, dual energy x ray absorptiometry
• FFM, fat free mass
• FM, fat mass
• GH, growth hormone
• NHL, non-Hodgkin’s lymphoma

## Statistics from Altmetric.com

Body weight is composed of two major compartments, fat free mass (FFM) and fat mass (FM), such that total body weight = FFM+FM. Although body weight is relatively straightforward to measure, estimation of FFM and FM is fraught with difficulties. Often they are measured indirectly with conversion of the measurement into compartmental body composition using regression equations and/or “constants”. For example, from measuring four peripheral skinfold thicknesses (biceps, triceps, subscapular, and suprailiac), FM may be estimated by using previously published equations.1,2 FFM may be estimated by measuring total body water and then making an assumption that FFM has a constant hydration state.3,4 Such equations and “constants” are derived from the relationship between a technique and an accepted “gold standard” method in normal healthy children. However, usage of such formulae or “constants” derived from the normal population may be unreliable, particularly when applied to children who have undergone therapies that may alter the relationships between these “constants” and body composition, and reliance on such formulae may lead to clinically significant errors.5

To test this hypothesis measures of body composition have been made in children previously treated for acute lymphoblastic leukaemia (ALL). Using indices of height and weight (such as body mass index (BMI): weight/height2) these children are recognised to become obese, and may have altered body composition for a variety of reasons, including endocrinopathies resulting from previous cranial irradiation or chemotherapy.6–12 However, the normal relationship between height and weight is altered in disease states and BMI is an inappropriate measure of body fat in such children.13 In this study the hydration and potassium content of FFM were measured using deuterated water dilution and whole body potassium counting. Dual energy x ray absorptiometry (DEXA) was assumed to be a “gold standard” for measurement of FFM. Results of children treated for ALL have been compared with a group of children treated for malignancies other than ALL, who received chemotherapy alone, and a group of healthy sibling controls. Bioelectrical impedance (BIA) measurements allow an estimation of total body water, but rely on a constant hydration state. Furthermore, formulae derived for the estimation of body water from BIA are analyser dependent. Therefore, the effect of the derived hydration constant for the control population on children treated for ALL and other malignancies and comparison of impedance measurements by two different analysers is also explored.

## METHODS

### Subjects

The subjects studied have been previously described elsewhere, and formed a cohort of patients undergoing investigation into the onset of obesity following treatment for ALL.14–16 Briefly patients had all received treatment for malignancy at the regional oncology referral centre for South Wales in Llandough Hospital and Community NHS Trust, Cardiff. At the time of study, all the children were in continuous remission and none was receiving treatment (such as growth hormone (GH) therapy) likely to have an influence on body composition. Children were divided into three groups.

#### (1) ALL group

This group had been treated for ALL on standard Medical Research Council protocols during the period 1979 to 1990 and included UKALL VI, VIII,17 and X.18 Treatment consisted of an induction block to induce remission followed by a central nervous system (CNS) prophylaxis phase consisting of cranial irradiation (18 or 24 Gy) and intrathecal methotrexate, followed by maintenance chemotherapy for two or three years. The later protocols included further intensification blocks of more intensive chemotherapy. Induction chemotherapy included the use of corticosteroids, vincristine, and asparaginase with the introduction of daunorubicin in UKALL VIII and X. Maintenance chemotherapy consisted of mini reinduction blocks of vincristine and corticosteroids with oral methotrexate and 6-mercaptopurine.

#### (2) Other malignancies group

This treated comparison group comprised children treated for a variety of other malignancies including acute myeloid leukaemia (AML; n = 7), non-Hodgkin’s lymphoma (NHL) (n = 2), Wilms’ tumour (n = 6), neuroblastoma (n = 2), yolk sac tumour (n = 2), rhabdomyosarcoma (n = 1), and Hodgkin’s lymphoma (n = 1) between 1982 and 1991. This group, as a whole, had been treated with cytotoxic chemotherapy but no cranial or other radiotherapy. Chemotherapy for this group was given in pulsed blocks, allowing bone marrow recovery between these blocks, as opposed to the more continuous nature of therapy for ALL. A variety of chemotherapeutic agents had been used, but in particular, apart from children treated for NHL or Hodgkin’s lymphoma (n = 3) this group had not received corticosteroid therapy.

#### (3) Sibling control group

For comparison with the two treatment groups, measurements were also performed on a healthy group of their siblings. Where there was a choice of more than one sibling within the appropriate age range, the one closest in age to the index case was chosen. Controls were siblings of both the treated groups and were not specifically age and gender matched. However, they fulfilled the age criteria for study and were free from any chronic disease and were not participating in sporting activities likely to affect body composition.

### Ethics committee approval

The study was approved by the ethics committee of the South Glamorgan Health Authority and written parental consent was obtained for each child.

### Body composition measurement

All auxological measures were made by the same trained observer (JTW) and were all made on the same day as the body composition measurements for each patient. Pubertal status was assessed for each child.

#### Dual energy x ray absorptiometry

DEXA measurements of body composition were made using the Hologic QDR 1000/W (Hologic, Inc., Waltham, MA, USA). This technique has been described elsewhere19 and is based on the attenuation characteristics of x rays of two different energies passing through the subject while lying supine. The technique is non-invasive and well accepted in children with the total radiation dose not exceeding 5 μSv (equivalent to one day’s background radiation), and a precision of better than 2%.20

#### Total body water

Total body water was measured from deuterium oxide (2H2O) dilution after administration of 0.3 g 2H2O/kg of estimated total body water by mouth after an overnight fast (2H was 99.9 atom% from Sigma Chemical Co.). Urine specimens were collected for analysis before and between three and five hours post dose. The 2H:H ratio was measured by continuous flow-isotope ratio mass spectrometry (Hydra system, Europa Scientific, Crewe, UK)21 and total body water was calculated from the 2H dilution method of Schoeller et al after standardising the enrichment of the given tracer by analysis of the suitable dilutions with tap water.4 FFM can be calculated from measurements of total body water assuming a constant hydration state of 73%; that is, FFM = total body water/0.73.

#### Whole body potassium

Total body potassium was determined using a whole body counting chamber. FFM can be calculated from total body potassium measurement either by the use of previously published equations or from constants, assuming a linear relationship between the total body potassium and FFM. The equation of Burmeister derives FFM from measurements of whole body potassium and body surface area using the formula:22

$Math$

#### Bioelectrical impedance

Bioelectrical impedance measurements were made using the RJL (RJL Systems Inc.) and Holtain (Holtain Ltd, UK) body composition analysers. The technique is simple to perform and highly acceptable for use in children. Electrodes were applied to the dorsum of the right wrist and to the flexor surface of the right ankle, through which a harmless 800 μA, 50 kHz high frequency alternating current was passed. The measurements were made consecutively using the same electrodes.

### Statistical analysis

Differences between groups were examined by analysis of variance (ANOVA) and if significant (p<0.05) the post hoc least significance test was applied to look for group differences. The Pearson correlation was applied to identify significant associations within the data, and the paired Student’s t test used to compare measures by two techniques. Regression analysis was used to produce formulae for the association between total body water and the impedance index. No statistical adjustment was made for any family effect of using siblings from the two treatment groups as controls. A complex statistical model would be required to rule out such an effect and such analysis would not be expected to alter the results or conclusions in any appreciable way. Statistical significance was taken as p<0.05. Analyses were performed using the Statistical Package for Social Sciences (SPSS) version 6.0.

## RESULTS

Table 1 presents patient details. There were no significant differences in age, height, and weight (and SD scores) between the three groups. Children treated for ALL were significantly fatter than those with other malignancies when height and weight were expressed as body mass index (BMI) and had significantly more fat than both other groups when FM was expressed as %fat derived from DEXA scanning (table 1). There was no significant difference in pubertal status between groups.

Table 1

Mean (SD) patient details for the three groups

### Body water

There were no significant differences in the total body water measurements made by isotopic water dilution between groups (table 1). However, the hydration state of FFM (total body water/FFM) using total body water measurements from isotopic dilution and FFM derived from DEXA showed a significant reduction in children treated for ALL compared to both the other malignancies and control groups (table 2).

Table 2

Mean (SD) hydration and 40K constant for fat free mass for the three groups

### Body potassium

There were no significant differences in the total body potassium between groups (table 1). However, the potassium content of FFM (total body potassium/FFM) was significantly reduced in children treated for ALL compared to the other two groups (table 2).

### Bioelectrical impedance

There were no significant differences between groups for impedance measurements from each analyser (table 1). However, despite a good correlation between the RJL and Holtain analysers (r = 0.98, p<0.001), the Holtain analyser gave significantly lower impedance readings compared to the RJL analyser (mean (SD) 488 (65) Ω versus 586 (84) Ω respectively, p<0.001, for all groups combined). There was a significant correlation in this discrepancy (r = 0.78, p<0.001) with greater differences between the two analysers for higher impedance values (that is, those with less FFM) (fig 1). Total body water measured by isotopic dilution was correlated with the impedance index (height2 (cm)/impedance (Ω)) measured by both analysers (fig 2).

Figure 1

Difference between impedance values from the RJL and Holtain analysers against the mean impedance for the two methods. r = 0.78, p<0.001. Dashed lines represent limits of agreement.33

Figure 2

Relationship between total body water measured by isotopic water dilution and the impedance index (height2/impedance) for the three groups combined. Squares represent measurements from Holtain, r = 0.98, p<0.001. Circles represent measurements from RJL, r = 0.98, p<0.001.

## DISCUSSION

This study describes significant reductions in hydration and potassium constants for FFM in children previously treated for ALL compared to children treated for a variety of other malignancies and healthy controls. Hence, previously published constants and equations which rely on such variables remaining constant, may lead to inappropriate conclusions about body composition in children treated for ALL. For example, using a recognised published hydration constant of 0.73 litres water/kg FFM,23 children treated for ALL would be predicted to have significantly lower estimates of FFM when calculated from total body water measurement than that derived from DEXA by a mean (95% CI) of −1.15 (−0.47 to −1.83) kg, p<0.01, with no significant differences for the other malignancies and control groups. Therefore application of a hydration constant derived from normal children will have implications for the estimation of FFM in children treated for ALL and will lead to underestimation of FFM and hence overestimation of FM. Likewise application of the potassium constant derived from controls (63.43 mmol/kg FFM) to the other two groups leads to an underestimation of FFM compared to that derived from DEXA by a mean (95% CI) of −2.29 (−1.15 to −3.43) kg for the ALL group, whereas there were no differences for the other malignancies group. Furthermore, comparison of FFM derived from the equation of Burmeister22 (see above) with that derived from DEXA revealed an underestimation of FFM derived from total body potassium by a mean (95% CI) of −1.65 (−0.54 to −2.76) kg, p<0.01 for the ALL group, but no significant differences for the other malignancies or control groups. Therefore assumption of a constant potassium state would lead to an overestimation of FM for the ALL group.

DEXA is a relatively new technique for body composition analysis. It requires little cooperation and the radiation exposure is extremely small. It provides a more direct analysis of body composition, comparing attenuation characteristics of x rays to known standards for estimation of fat, lean, and bone mineral mass and is relatively insensitive to changes in the hydration state of lean tissue and is therefore suitable for use in children. It is now widely accepted as a “gold standard” for body composition measurement in adults and children.19

The reduction in hydration and potassium constants (table 2) in children treated for ALL implies a depleted hydration and potassium state of FFM. The constants for the control group compare favourably with previously published data suggesting that the DEXA measurement of FFM in the control group is accurate.

The mechanism underlying the relative dehydration in children treated for ALL is not clear. The finding of a normal hydration state in children treated for a variety of other malignancies suggests that the under-hydration state observed following ALL is specific for that disease process or its therapy. Under-hydration of FFM may result from an endocrinopathy such as GH deficiency, as has been shown in adults.24,25 Although GH deficiency is recognised following cranial irradiation, it is related to the dose received.26 Low dose cranial irradiation, such as that used to treat the children with ALL in the current cohort, is not normally associated with GH deficiency27 but has been associated with subtle abnormalities of GH pulsatility.28 Measurements of GH status were not made in the children participating in the current study since they were growing normally at the time. Furthermore, making a diagnosis of GH deficiency in a normally growing overweight child is difficult. Hence we can only hypothesise that subtle abnormalities of GH secretion may lead to abnormalities of hydration state albeit not severe enough to cause significant abnormalities of growth. Alternatively, FFM may have been overestimated by DEXA due to an effect of the increased levels of FM in children treated for ALL on the measurement technique leading to a reduction in the derived hydration state. Although no in vivo data exist to support this hypothesis, in vitro data using pork carcasses has shown that DEXA tends to underestimate FM and that the absolute error increases as the proportion of fat increases.29 Therefore, for fatter individuals body fat may be relatively underestimated by DEXA leading to an overestimation of FFM and hence an artefactual lower hydration state.

In the current study, use of previously published equations, or constants for the total body potassium content of FFM derived from healthy children led to underestimates of FFM in children treated for ALL when estimated from total body potassium counting. This appears to be due to a relative potassium depletion of FFM in children treated for ALL (table 2). GH deficient adults have previously been shown to have a lower potassium content of FFM.30 It may therefore be postulated, as for total body water, that subtle abnormalities of GH pulsatility may be responsible for the potassium deplete state seen in children treated for ALL. Alternatively, the increased body fat in children treated for ALL may cause radiation attenuation which is not completely corrected by the method previously described, leading to an under estimation in the true total body potassium and hence the potassium content of FFM.

Bioelectrical impedance is often used for measurement of FFM in children because it is non-invasive, simple to use, and well tolerated. However, as shown in this report and by others,31,32 care must be taken in the choice of predictive regression equations used for the calculation of total body water and FFM. In the current study the regression equations for the relationship between total body water and the impedance index derived from controls was significantly different for the two analysers:

$Math$

$Math$

Application of these equations for the calculation of total body water from impedance measurements significantly underestimated the measured value from isotopic water dilution in children treated for ALL when performed on the RJL analyser (mean (95% CI) difference: −0.74 (−0.21 to −1.27) litres, p<0.01), but not on the Holtain analyser (mean (95% CI) difference: −0.61 (+0.13 to −1.35) litres, p = 0.10). There were no differences for the estimation of total body water from impedance for either analyser when compared to the value obtained from isotopic water dilution for the other malignancies group.

Using the predictive formulae for total body water from the Holtain and RJL analysers from the current study, a normal child of 150 cm with an impedance of 550 Ω would be predicted to have a body water of 20.9 litres and 25.0 litres respectively. Assuming a hydration state of 73% for FFM this would give FFM values of 28.6 kg and 34.2 kg respectively. Recently four different published equations for the prediction of FFM from bioelectrical impedance were applied to measurements made in 98 prepubertal children.31 Wide limits of agreement were found between the different equations and it was concluded that the general applicability of impedance equations cannot be safely assumed. It is therefore recommended that body composition should be measured by more than one methodology to allow cross validation. Alternatively, predictive formulae for total body water from impedance should be established in one’s own centre using isotopic water dilution as the “gold standard”.

In conclusion, children treated for ALL have a reduced water and potassium content of FFM. Care should be taken in the application of constants or regression equations derived from healthy individuals in the assessment of body composition in children with disease states such as ALL.

## Acknowledgments

We are grateful to Mr JH Pearce, Mrs RJ Pettit, and Mr D Coleman (Department of Medical Physics, University Hospital of Wales) for performing the DEXA and total body potassium scans. We also wish to thank Europa Scientific Limited, Crewe, for performing the isotopic ratio mass spectrometry, and in particular to Dr S Brooks and Mr AP Bradley for guidance with the preparation of the samples for analysis. Finally we thank the children and their parents for participating and making the study possible.

View Abstract

## Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.