Dual-Energy X-Ray Absorptiometry and Anthropometry for Assessment of Nutritional Status at Diagnosis in Children with Cancer: A Single-Center Experience from India

Introduction

The survival of children with cancer has improved substantially over the past few decades. This has been possible owing to improvements in surgical and radiation techniques, use of risk and response-based treatment approaches to chemotherapy, and better supportive care. Undernutrition and low socioeconomic status have been reported to be associated with higher treatment-related toxicity, abandonment of therapy, and poor overall survival.1 Hence, nutritional assessment and intervention are now recognized to be important components of supportive care in the management of children with cancer.

It is estimated that at least 80% of children with cancer live in low- and middle-income countries (LMICs) and the average prevalence of undernutrition in this population is 50% (30–80%).2 Various tools have been used for the assessment of nutritional status, of which anthropometry is the most common and inexpensive method. However, due to the use of a wide range of anthropometric indices it is often difficult to make meaningful comparisons. Other methods of assessing nutritional status include biochemical tests and dual-energy X-ray absorptiometry (DXA). The latter has the additional advantage of being a widely accepted and precise tool for measuring body composition.3 DXA has been used most often in children with leukemia but there are scant published data from LMICs.

Hence, the present investigation was undertaken to study the nutritional status of newly diagnosed children with cancer using DXA, anthropometry, and biochemistry, and to compare them using lean body mass (LBM) by DXA as the criterion measure of nutritional status. In addition, the impact of nutritional status on treatment-related morbidity during the first 3 months of therapy was evaluated.

Material and Methods

Results

The characteristics of 137 newly diagnosed pediatric cancer patients are presented in Table 1. Their mean age was 10.5 (±4.2) years and the M:F gender ratio was 4.7:1. The majority of patients belonged to the low and middle socioeconomic strata (96%). Anthropometric assessments and serum albumin were available for all patients and DXA for 101 of them (age > 3 years). History of weight loss was present in 29% of them, median duration of illness was 54 days, 55% had hematological malignancies, and 25% had metastatic or high-risk disease at diagnosis. The mean values for BMI (z-score –0.617), MUAC (5th percentile), TSFT (50–75th percentile), LBM (z-score –1.399), FM (z-score 0.446), and serum albumin (3.9 g/dL) were calculated (not shown in Table 1).

The prevalence of undernutrition and obesity by the different measures is depicted in Table 2. The various anthropometric tools detected undernutrition among 14% (BMI), 23% (TSFT), 45% (MUAC), and 64% (AMC) of the patients. Serum albumin levels detected 25% patients to be undernourished and the prevalence was higher among those with HM than in those with ST (p, 0.008). Inclusion of serum albumin to arm anthropometry in the categorization resulted in a final distribution of adequately nourished 32%, moderately depleted 18%, and severely depleted 49%.

For analysis of body composition by DXA application of LBM z-score < –2 and < –1.645 (equivalent to 5th percentile) revealed 33 and 44% patients to be undernourished (or sarcopenic), respectively. Using FM by DXA, all patients had z-scores between –1.645 and +2 and none had low FM. Further analysis revealed that among patients detected to be adequately nourished by MUAC, 25% (13/51) were sarcopenic by LBM (Supplementary Table S1, available online only).

To study the relationship between the various tools a correlation analysis was performed (Table 3). The anthropometric parameters MUAC (r = 0.68) and BMI (r = 0.82) correlated well with LBM while TSFT correlated well with FM (r = 0.86) with a significant p-value < 0.0001 (Supplemental Table S1, available online only). Correlations were considered to be statistically significant by taking Bonferroni adjustment into account.

Univariate and multivariate linear regression analyses were done to compare the various nutrition parameters using LBM by DXA as the criterion measure. It was observed that MUAC (R² = 0.681) was the best predictor of undernutrition with the lowest standard error (Table 3). The multivariate regression analysis shows that the two arm anthropometric parameters, MUAC and TSFT, accounted for 78.6% of the variation in LBM.

Overall patients with HM had significantly more morbidities than those with ST, irrespective of their nutritional status (Table 4). The undernourished and adequately nourished patients experienced similar treatment-related toxicities in both groups. Among patients with hypoalbuminemia a higher proportion were observed to have febrile neutropenia (p, 0.039), chemotherapy delays (p = 0.003), and interval admissions (p = 0.025). No association was found between nutritional status and socioeconomic strata.

Discussion

The nutritional status of children with cancer is extremely important, since a good nutritional status enables them to cope better with cancer therapy and allows us to administer intensive treatments often required for curative strategies. Better understanding of the burden of undernutrition and how it affects the different body compartments is also relevant as it helps develop the foundation for nutrition intervention. This is a single-center study from a tertiary care cancer center from North India that caters to patients from the middle and low socioeconomic strata of society. It is a challenge to compare data from other studies as several factors influence the interpretation of nutrition data, the most compelling being the use of different anthropometric tools in the assessment of nutritional status and lack of uniformity in defining cutoffs for undernutrition. There are very few investigators who have used DXA in newly diagnosed pediatric oncology patients and none from India. Moreover, while comparing nutrition data between patients from high-income countries and LMICs it is important to consider differences in body composition that may be attributable to differences in ethnicity, environmental, and dietary factors. Hence, the authors proposed the use of LBM by DXA along with other anthropometric indices to offset this ambiguity.

In the present study, nutritional evaluation by arm anthropometry (MUAC) and DXA (LBM) revealed that approximately two-fifths of the patients were undernourished while overweight and obesity were quite rare at the time of diagnosis. The observed mean values for anthropometric indices: BMI z-score, MUAC, and TSFT were higher than those reported by studies from two large centers in India.1213 The incidence of undernutrition reported in these studies was also higher (70–90%) and probably attributable to the fact that these centers cater to patients from relatively lower socioeconomic strata of society.1213 Internationally, the results of the present study match closely with the reported prevalence in other LMIC, that is, from Central America and Brazil.914

The findings of our study revealed BMI to be an imprecise measure of nutritional status especially at diagnosis, and compared with LBM and MUAC its use misclassified nearly 30% of the patients as normally nourished. Anthropometric indices based on weight and height often underestimate undernutrition as they are unable to account for tumor-related factors like weight of tumor, edema, etc., especially in children presenting with advanced and large volume disease. Body composition analyses and arm anthropometry on the other hand remain unaffected to a significant extent by tumor mass and thus are better able to assess the nutritional status in children with cancer. Similar observations have been made by others in India and abroad.121516171819202122

DXA is a reliable tool for measurement of body composition in clinical practice despite its limited availability outside of resource-rich settings. There is little published data on use of LBM by DXA to assess nutrition status in pediatric oncology patients at diagnosis182324 from LMICs. In the present study, LBM detected a higher prevalence of undernutrition (44%) when compared with that detected by BMI (14%). We observed a good concordance between LBM and MUAC in detecting undernutrition. A strong correlation of LBM was seen with MUAC (and AMC), while FM correlated well with TSFT (Table 3). A study on Canadian patients by Barr et al reported a similar correlation for LBM and MUAC but observed a weaker correlation between TSFT and FM.18 A previous report from Mexico23 conducted among patients with acute lymphoblastic leukemia, DXA detected more undernourished patients than BMI (20% vs. 11.8%), although correlation of the various parameters with each other was not described. Adopting LBM as the clinical gold standard, we found MUAC to have the best value in prediction of undernutrition among the various anthropometric indices.

Interestingly, in the present analysis, LBM and MUAC detected undernutrition among 44 to 45% of patients while TSFT detected this in just 16% of patients. Sarcopenia or skeletal muscle depletion with or without fat loss has been shown to be a hallmark of cancer cachexia.22 One may postulate that in our undernourished population arm muscle is lost before arm fat which is preserved till the very last.

It is noteworthy that none of our patients had low FM z-score (< –1.645), in other words FM as detected by DXA was preserved among all patients, even those observed to be undernourished by other parameters (MUAC, TSFT, and LBM). Furthermore, at times of acute stress/disease (cancer) central fat (as measured by FM) is relatively preserved compared with peripheral fat (as measured by TSFT).25 This associated with mean high FM (z-score 0.44) and TSFT (50th–75th centile) at diagnosis would appear to indicate that our patient population have a high baseline arm and whole body fat.

We also observed that more than a quarter (28%) of our adequately nourished patients (by MUAC) was sarcopenic (by LBM). These findings concur with published reports on ethnic variations in body composition showing that for a given BMI, South Asian populations have lower LBM and higher visceral fat compared with their western counterparts.26

These findings on body composition, that is, high body fat and low muscle mass, have important implications on interventions, in terms of type of dietary and physical activity recommendations to our cancer patients.

We studied the incidence of various treatment-related morbidities during the first 3 months of cancer therapy, and found all toxicities to occur more frequently among patients with HM than those with ST (p = 0.01). Although patients with HM had a higher prevalence of undernutrition, there was no difference in morbidities between the undernourished and adequately nourished children either by anthropometry or DXA. This is in contrast to other studies which found more episodes of febrile neutropenia and delays in chemotherapy among undernourished patients.1527 We believe this to be probably due to heterogeneity in distribution of cancer type, stage/risk group, and intensity of treatment in the two nutrition categories. We observed hypoalbuminemia to occur in one-fourth of the patients and low albumin levels at diagnosis correlated with higher incidence of febrile neutropenia, chemotherapy delays, and longer interval admissions (Table 4). It is well known that low serum albumin is an excellent marker of metabolic stress due to various reasons including cancer and infections in addition to malnutrition. It has been previously shown to be a predictor of both morbidity and mortality, and hence clinicians must keep this in mind while planning treatment and supportive care for patients with hypoalbuminemia. In fact, investigators have combined arm anthropometry with serum albumin in nutrition intervention algorithms to predict patients at high risk.910

The strengths of our study are that it was designed prospectively and appears to be the only one from an LMIC that evaluated nutritional status and body composition using DXA and compared it with various anthropometric indices. Moreover, we recorded and analyzed the morbidities among all the patients and correlated them with cancer type and nutritional status. The limitations include its relatively small sample size and skewed gender ratio in favor of males as also reported from other studies in India.28 Although dietary assessment and interventions were done, these were not systematically planned or done sequentially. Indian reference values for DXA parameters were not available for our population.

Conclusion

Our study reports undernutrition among two-fifths of newly diagnosed patients of childhood cancer belonging to low- and middle-income socioeconomic strata, as detected by arm anthropometry and corroborated using LBM by DXA as the criterion measure. Body composition analyses revealed our patients to be more sarcopenic and adipose even when adequately nourished by anthropometry. These findings are extremely relevant in planning intervention strategies for newly diagnosed patients. Larger studies having a more homogenous cohort of patients, with serial longitudinal nutritional status assessments may help determine the influence of nutritional status at diagnosis on the final clinical outcome.