Reilly, John J.*; Weir, Jennifer*†; McColl, John H.†; Gibson, Brenda E. S.‡
There is a perception that undernutrition is common in children with cancer and that it is often present early in the course of the disease (1,2). Undernutrition in childhood cancer is therefore a general concern (2-4) because of its functional consequences, but in acute lymphoblastic leukemia (ALL), the most common malignant disease in childhood, there is also specific evidence that suggests that protein-energy undernutrition at diagnosis is an adverse prognostic factor (5-7), although it may be of limited relevance in developed countries (8).
Despite the size of the patient population and the potential clinical significance of undernutrition in ALL, there is no consensus on the question of whether undernutrition is common early in therapy in patients with ALL. Some investigators have concluded that it is common (2,9,10), others that it is not (1,11,12). This is probably a consequence of relatively small sample sizes in previous studies (1,9,12,13) and differences in methods (13,14). Some have defined nutritional status at diagnosis on a biochemical basis (1,9), whereas others have used anthropometry but with a variety of different indices and criteria for defining undernutrition (1,11-13).
The purpose of the present study was to test the hypothesis that undernutrition is common by describing nutritional status at diagnosis in a large, national cohort of patients treated for ALL, obviating concerns about sample size and patient selection. There is currently no gold standard for pediatric nutritional assessment (14,15), but a simple and objective anthropometric assessment of protein-energy status is provided by the body mass index (BMI; body weight in kilograms/height in meters2). There is now a consensus that undernutrition can only be reliably detected in children if such an anthropometric approach to nutritional assessment is adopted (15-18). The BMI is a good proxy for body composition (body fat and lean body mass) in childhood (19) and is therefore an index of protein-energy depletion or excess (15-18). In addition, contemporary reference data are available (20,21) for the BMI, and it is therefore widely recommended for nutritional assessment of children (18,20-22). The BMI is attractive from a practical point of view in pediatric oncology because it depends on simple measurements (weight and height) that are routinely made to calculate drug dosage. An additional advantage of defining undernutrition on the basis of the BMI is that the alternatives to assessment of protein-energy status in childhood cancer have serious limitations. In particular, interpretation of biochemical or haematologic markers of nutritional status early in malignant disease is confounded by the underlying disease (13,23). We therefore used the BMI as the index of protein-energy undernutrition in the present study and expressed BMI as standardised scores (SDS) relative to United Kingdom reference data (20), as currently recommended (18,20-22,24).
Data were obtained retrospectively on all 1033 patients treated for standard-risk ALL in the United Kingdom on Medical Research Council protocol UKALL-X (1986-1991). Eight patients were excluded from the analysis because their conditions had been incorrectly diagnosed. A further six patients were excluded because their heights and/or weights had not been recorded at diagnosis, leaving 1019 patients in the analysis.
Assessment of Nutritional Status
In each patient weight (to 0.1 kg) and height (to 0.1 cm) are measured at diagnosis for calculation of drug dosages. We extracted these data, calculated BMI, then expressed BMI as a standardised age-and sex-independent SDS, relative to UK population reference data (20), using a software package (Child Growth Foundation, London, UK). This value summarises each BMI in terms of the number of SD units above or below the mean of the reference population. Undernutrition was defined as BMI SDS less than -2.0 (between the 2nd and 3rd percentiles) and overnutrition as BMI SDS more than 2.0 (between the 97th and 98th percentiles) (15,17-22). Differences between observed and expected frequencies of children below BMI SDS -2.0 and above BMI SDS +2.0 were tested for significance using the χ2 goodness-of-fit test. Two sided t-tests were used to test the null hypothesis that mean BMI SDS of the sample equalled zero.
Because both the mean BMI SDS and the variance in BMI SDS were likely to have an influence on the apparent prevalence of undernutrition, it was appropriate to examine the underlying assumption of the process of calculating BMI SDS values (i.e., that variance of the BMI SDS values = 1.0) by testing whether this was the case using two-sided χ2 tests.
Median age at diagnosis was 4.4 years (range, 0.4-14.9 years) in boys and 4.4 years (range, 0.2-14.9 years) in girls. Mean BMI SDS values at diagnosis were consistently negative in both boys (Table 1) and girls (Table 2). For the entire sample of boys, mean BMI SD score was -0.32 (SD 1.30; 95% CI -0.41 to -0.23). For the entire sample of girls, mean BMI SD score was -0.28 (SD 1.20; 95% CI -0.43 to -0.13). The observed frequency of children with BMI SDS less than -2.0 was 58 (7.6%) of 767 in boys and 17 (6.7%) of 252 in girls, compared with an expected frequency of 2.3% in each case (16,19,20). Differences between these observed and expected frequencies were statistically significant (p < 0.0001). Overall, the 95% CI for percentage of children with BMI SDS less than -2.0 was 5.8% to 9.0%. This indicates a threefold excess of undernutrition at diagnosis in both sexes (14,17,19,20,22).
The frequency of overnutrition (BMI SDS <2.0) (14,17,20,21) was 20 (2.6%) of 767 in boys and 8 (3.2%) of 252 in girls. These differences between observed and expected frequencies for overnutrition at diagnosis were not statistically significant.
Overall, in both boys and girls, there was strong evidence that variance of the BMI SD scores differed significantly from 1.0 (p < 0.001 for both sexes). Levene's test for the homogeneity of variances across age groups returned p = 0.051 in boys and p = 0.612 in girls. The resultant pooled estimate of the standard deviation was 1.301 for boys and 1.211 in girls.
This study provides the first data on nutritional status at diagnosis in a national cohort of patients with ALL. The observation that mean BMI SD scores of both boys and girls were negative, and the higher than expected prevalence of children below BMI SDS of -2.0, suggest that nutritional depletion is present in this patient population at diagnosis. This finding applies across the age range (Tables 1 and 2) and is an argument for routine nutritional screening at diagnosis in ALL. Because height and weight are routinely measured and recorded at diagnosis, the BMI could form the basis of standardised nutritional screening at diagnosis in childhood leukemia, and it is an index that now has widespread acceptance (17-22).
Although the BMI SDS has its limitations, it is a useful clinical index of protein-energy malnutrition (and excess, i.e., obesity) that is simple to calculate and interpret and is inexpensive (15,18-22). Biochemical assessment of protein-energy status at diagnosis in children with leukemia can be misleading (13,23) and is neither simple to interpret nor inexpensive (15,16,23). Objective, anthropometric, assessment of nutritional status is essential if valid assessments are to be made in pediatrics (16).
Choice of BMI, rather than an alternative anthropometric index (such as midarm circumference or triceps skinfold), may depend on the circumstances. In children with a large solid tumour burden or oedema, nutritional depletion may be masked by using indices based solely on body mass, such as BMI or weight for height (13,24), and measurement of indices such as midarm circumference (an index of muscle mass), with or without triceps skinfold (an index of fat mass), may therefore be necessary. However, the BMI has advantages over these other forms of anthropometry in that it is simpler and therefore more suitable in the clinical setting, and the interpretation of the index is relatively simple. It is also worth noting that current UK population reference data for midarm anthropometry are unsuitable for young children (25) and, given the greater degree of practical difficulties, midarm anthropometry may not be the preferred option for routine nutritional screening in patients with ALL.
Using the BMI to screen for undernutrition requires a degree of confidence in the underlying reference data. In the UK, the first version of the 1990 reference data for infants was criticised (26), but the revised version appears to be adequate (25,27). Evidence in the present study that variance of BMI SD scores significantly exceeded 1.0 would have been a concern if we had studied a population of healthy children. However, this apparent violation of one of the underlying assumptions of the BMI reference data may have reflected the effect of the disease on nutritional status. One final caveat in relation to use of BMI to screen for undernutrition concerns accuracy of the basic measurements of weight and height. The size of the cohort we studied would have minimised the effect of measurement error and so would not have influenced our conclusions. However, when making clinical judgments based on BMI in individual patients the accuracy of weight and height measures should be established.
Quantitative comparisons between prevalence of undernutrition in this cohort of children with ALL and other smaller studies are difficult to make because of differences in methods used. In addition, in all previous studies much smaller and more selected samples of children with ALL were recruited (in many cases with sample sizes <20). Several investigators (1,11,12) reported no significant differences in anthropometric indices between patients recently diagnosed with leukemia and controls and implied that prevalence of undernutrition at diagnosis was relatively low. However, Yu et al., (9) in a small study of children with ALL at various stages of treatment, found that biochemical indices of nutritional status were generally low, but these may have been lowered artifactually by any acute-phase response.
There is a widespread perception that undernutrition may be common early in the course of childhood leukemia (1,2) and that this may be secondary to the anorexia that is not uncommon before diagnosis. However, as noted earlier, there was some evidence that suggested that nutritional depletion is uncommon early in ALL. A larger study of nutritional status using methods that were not readily confounded by the acute phase response was necessary. The size of the present study, together with the use of contemporary anthropometric reference data, provides the strongest evidence to date that undernutrition is a feature of early ALL. Within the cohort there was no evidence that undernutrition was related to other clinical features such as white cell count at diagnosis or age of the patient, and therefore, these could not be readily used as proxy markers for undernutrition.
Although there is no gold-standard definition of undernutrition in childhood (13,14), we conclude that undernutrition at diagnosis in ALL is approximately three times higher than expected. This constitutes an argument for formal objective screening based on anthropometry. If such screening is not adopted, a substantial number of undernourished patients will not be identified as such (16). Undernutrition could be detected by use of information (weight and height) routinely collected at diagnosis.
Protein-energy malnutrition has adverse clinical implications, including impaired tolerance of chemotherapy and impaired immune function (2,3,6,7). For most (but probably not all) patients in developed countries, undernutrition tends to resolve and is not associated with increased risk of relapse in the long term (8). However, undernutrition may be associated with other adverse outcomes in the short term, and in developing countries, severe undernutrition is more common and therefore more of a concern (6,7).
If undernutrition can be detected relatively simply using the BMI and treated by referral to the hospital nutrition team or dietetic department, the case for nutritional screening at the time of diagnosis seems strong. In the medium to long term, the major nutritional risk for these patients remains overnutrition (obesity) (28-30), and any screening programme based on weight and height measurements would be suitable for identification of both under- and overnutrition in childhood leukemia. Simple clinical guidelines for the identification and management of overweight and obesity based on the BMI are available (31). The purpose of nutritional screening would therefore be to identify those patients who may require specialist nutritional intervention and more appropriate referral to such services.
Acknowledgment: The authors thank Diane Henderson, trials coordinator, Yorkhill Hospitals, for her assistance; Professor O. B. Eden, Chairman, MRC Childhood Leukaemia Working Party, and Dr. Sue Richards and Julie Burrett, Clinical Trials Service Unit Oxford, for their assistance and guidance; and members of the Childhood Leukaemia Working Party for allowing data from their patients to be used in the research.
The study was supported by the Leukaemia Research Fund.
1. Delbecque-Broussard L, Gottrand F, Ategbo S, et al. Nutritional status of children with ALL: A longitudinal study. Am J Clin Nutr
2. Donaldson SS, Wesley MM, De Wys WD, Suskind RM, Jaffe N, Van Eys J. A study of the nutritional status of pediatric cancer patients. Am J Dis Child
3. Van Eys J. Effect of nutritional status on response to therapy. Cancer Res
4. Reilly JJ, Workman P. Is body composition an important variable in the pharmacokinetics of anticancer drugs? Cancer Chemother Pharmacol
5. Reilly JJ, Odame I, McColl J, McAllister PJ, Gibson BES, Wharton BA. Does weight for height have prognostic significance in children with ALL? Am J Pediatr Hematol Oncol
6. Borato-Viana M, Murao M, Ramos G, et al. Malnutrition as a prognostic factor in lymphoblastic leukaemia: A multivariate analysis. Arch Dis Child
7. Lobato-Mendizabal E, Ruiz-Arguelles GJ, Marin-Lopez A. Leukaemia and nutrition I. Leuk Res
8. Weir J, Reilly JJ, McColl JH, Gibson BES. No evidence for an effect of nutritional status at diagnosis on prognosis in children with acute lymphoblastic leukemia. J Pediatr Hematol Oncol
9. Yu LC, Kuvibidila S, Ducos R, Warrier RP. Nutritional status of children with leukaemia. Med Pediatr Oncol
10. Smith DE, Stevens MCG, Booth IW. Malnutrition at diagnosis of malignancy in childhood: Common, but mostly missed. Eur J Pediatr
11. Uderzo C, Rovelli A, Bonomi M, et al. Nutritional status in untreated children with acute leukemia as compared with children without malignancy. J Pediatr Gastroenterol Nutr
12. Tamminga RYJ, Kamps WA, Drayer NM, Humphrey GB. Longitudinal anthropometric study in children with ALL. Acta Paediatr
13. Brennan BMD, Thomas AG. Nutritional status in children with acute leukemia. J Pediatr Gastroenterol Nutr
14. Wright JA, Ashenburg CA, Whitaker RC. Comparison of methods to categorise undernutrition in children. J Pediatr
15. Reilly JJ. Nutritional assessment. In: Campbell AGM, McIntosh N, eds. Fofar and Arneil's textbook of pediatrics
. 5th ed. Edinburgh: Churchill Livingstone 1997:1186-8.
16. Cross JH, Holden C, McDonald A, Booth IW. Clinical examination compared with anthropometry in evaluating nutritional status. Arch Dis Child
17. Cole TJ. Weight-stature indices to measure underweight, overweight, and obesity. In: Himes JH, ed. Anthropometric assessment of nutritional status
. New York: John Wiley & Sons, 1991:83-112.
18. Infants and children. In: Physical status: Use and interpretation of anthropometry
. WHO Technical Report Series 854. Geneva: World Health Organization, 1995:161-262.
19. Pietrobelli A, Faith MS, Allison DB, Gallagher D, Chiumello G, Heymsfield SB. Body mass index as a measure of a diposity among children and adolescents: A validation study. J Pediatr
20. Cole TJ, Freeman JV, Preece MA. Body mass index reference curves for the UK, 1990. Arch Dis Child
21. Rosner B, Prineas R, Loggie J, Daniels BR. Percentiles for BMI in US children 5-17 years of age. J Pediatr
22. Dietz WH. Use of the BMI as a measure of overweight in children and adolescents. J Pediatr
23. Talwar D, Ha TKK, Scott HR, et al. Effect of inflammation on measures of antioxidant status in patients with non small cell lung cancer. Am J Clin Nutr
24. Smith DE, Booth IW. Nutritional assessment of children: Guidelines on collecting and interpreting anthropometric data. J Hum Nutr Diet
25. Savage SA, Reilly JJ, Edwards CA. Adequacy of standards for assessment of growth and nutritional status in infancy and early childhood. Arch Dis Child
26. Wright CM, Corbett SS, Drewett RF. Sex differences in weight in infancy and the UK 1990 growth standards. BMJ
27. Preece MA, Freeman JV, Cole TJ. Sex differences in weight in infancy: Published centile charts for weight have been updated [letter]. BMJ
28. Odame I, Reilly JJ, Gibson BES, Donaldson MCD. Patterns of obesity in boys and girls after treatment for acute lymphoblastic leukaemia. Arch Dis Child
29. Van Dongen-Melman JEWM, Hokken-Koelega ACS, Hahlen K, De Groot A, Tromp CG, Egeler RM. Obesity after successful treatment of acute lymphoblastic leukemia in childhood. Pediatr Res
30. Didi M, Didcock E, Davies HA, Ogilvy-Stuart AL, Wales JKH, Shalet SM. High incidence of obesity in young adults after treatment of acute lymphoblastic leukemia in childhood. J Pediatr
31. Barlow SE, Dietz WH. Obesity evaluation and treatment: Expert committee recommendations. Pediatrics
© 1999 Lippincott Williams & Wilkins, Inc.