Children, cancer, and nutrition—A dynamic triangle in review

Authors

  • Alessandra Sala M.D.,

    1. Service of Hematology-Oncology, McMaster Children's Hospital, Hamilton Health Sciences, and McMaster University, Hamilton, Ontario, Canada
    Search for more papers by this author
  • Paul Pencharz M.B., Ch.B., Ph.D.,

    1. Division of Gastroenterology and Nutrition, Hospital for Sick Children and University of Toronto, Toronto, Ontario, Canada
    Search for more papers by this author
  • Ronald D. Barr M.B., Ch.B., M.D.

    Corresponding author
    1. Service of Hematology-Oncology, McMaster Children's Hospital, Hamilton Health Sciences, and McMaster University, Hamilton, Ontario, Canada
    • Department of Pediatrics, McMaster University, 1200 Main Street West, Hamilton, ON L8S 4J9, Canada
    Search for more papers by this author
    • Fax: (905) 521-1703


Abstract

The overall cure rate for cancer in childhood now exceeds 70% and is projected to reach 85% by the year 2010 in industrialized countries. Therefore, major attention is being placed on reducing the side effects of therapy. However, 85% of the world's children live in developing countries, where access to adequate care often is limited and health status frequently is influenced adversely by prevalent infectious diseases and malnutrition. Despite several confounding factors (different definitions of nutritional status, the wide variety of measures used for its assessment, the selection biases by disease and stage, treatment protocols of variable dose intensity and efficacy, small sample sizes of the studies conducted in the last 20 years), it is accepted that the prevalence of malnutrition at diagnosis averages 50% in children with cancer in developing countries; whereas, in industrialized countries, it is related to the type of tumor and the extent of the disease, ranging from < 10% in patients with standard-risk acute lymphoblastic leukemia to 50% in patients with advanced neuroblastoma. The importance of nutritional status in children with cancer is related to its possible influence on the course of the disease and survival. Some authors have described decreased tolerance of chemotherapy associated with altered metabolism of antineoplastic drugs, increased infection rates, and poor clinical outcome in malnourished children. In this article, the authors review methods of nutritional assessment and the pathogenesis of nutritional morbidity in children with cancer as well as correlations of nutritional status with diagnosis, treatment, and outcome. Cancer 2004;100:677–87. © 2004 American Cancer Society.

Cancer is the most common cause of disease-related death in children in industrialized societies,1 yet advances in the modalities of surgery, radiotherapy, and chemotherapy have led to major improvements in survival durations and cure rates for many types of malignant diseases in childhood. The overall cure rate now exceeds 70% and is projected to reach 85% by the year 2010.2 It has been estimated that, early in the new millennium, 1 in every 1000 young adults between the ages of 20–29 years will have been treated for cancer in early life.3 However, 85% of the world's children live in developing countries, where access to adequate care often is limited, delay in diagnosis is common, and health status frequently is influenced adversely by prevalent infectious diseases and malnutrition.4 In countries with limited resources, it is accepted widely that the prevalence of malnutrition averages 50% in children with cancer4; whereas, in industrialized countries, the prevalence of malnutrition is related to the type of tumor and the extent of the disease5 and is particularly common in patients with advanced neuroblastoma, Wilms tumor, and Ewing sarcoma.6

Malnutrition, notably protein/energy malnutrition (PEM), can affect tolerance of therapy, increase the risk of comorbidities, and influence the overall survival. Surprisingly, although numerous relevant articles on these issues have been published in the past 15–20 years, to our knowledge there are few if any comprehensive reviews, although the proceedings of an international workshop held in Mexico in 1997 have been reported.7

Conversely, but of escalating importance, the prevalence rates of being overweight and obesity continue to rise in children in the general populations of industrialized societies8 and some developing countries.9 In addition, there are increasing reports of obesity in children with cancer even at diagnosis and in the developing country setting.10

In this article, we have reviewed the nutritional assessment and the pathogenesis of nutritional morbidity in children with cancer as well as correlations with diagnosis, treatment, and outcome. In the context of pediatric oncology, the majority of studies on nutritional status have been conducted in children with acute lymphoblastic leukemia (ALL). There is an obvious need to obtain much more information in children with solid tumors who often experience a greater burden of nutritional morbidity related to both the biology of these diseases and the intensity of their treatment.

Assessment of Nutritional Status

To define nutritional status, weight and height usually are plotted on standardized growth charts and then converted to Z-scores. Z-scores are a numerical summary of each child's placement relative to an appropriate reference population, which is provided by the calculation of a score that is corrected for differences in weight and height between age-specific and gender-specific groups. Growth charts for North American children have been updated recently by the Centers for Disease Control and Prevention, but these charts have very limited applicability to children in developing countries, where such normative data seldom are available.

Any child whose height or weight Z-score is more than 2 standard deviations (SD) greater than or less than the appropriate mean is considered to have significant growth abnormalities. This allows the degree of acute and chronic malnutrition (and obesity) to be defined, as assessed clinically using the Waterlow criteria.11, 12 It is assumed that, during periods of nutritional deprivation, a weight deficit is the first abnormality to be noted, followed by a length or height deficit. The Waterlow criteria also assume that the expected height and weight measurements of a child follow the 50th percentile on the growth curves.

Acute malnutrition is defined as follows: weight deficit (%) = actual weight (kg) × 100/expected weight (kg) for actual height. Chronic malnutrition is defined as follows: height deficit (%) = actual height (cm) × 100/expected height (cm) at the 50th percentile for chronological age.

A child is considered to have first-degree, second-degree, or third-degree acute malnutrition if the weight-for-height estimate is < 90%, < 80%, or < 70% of the expected value, respectively. Likewise, a child is considered to have first-degree, second-degree, or third-degree chronic malnutrition if the height-for-age estimate is < 95%, < 85%, or < 75% of the expected value, respectively.

In addition, it is possible to calculate the body mass index (BMI) and the weight-for-height (WFH) ratio. BMI (weight/height squared; kg/m2) is a reasonably good proxy for body composition.13 The interpretation of BMI in childhood requires comparison with an appropriate reference population.14 Using this criterion, obesity is defined as a BMI > the 95th percentile.

However, there is no consensus on cut-off values to define malnutrition in adults, let alone in children. Although the WFH ratio is recommended by the World Health Organization,15, 16 it is potentially misleading in children with malignancy, especially in children with large solid tumors, and particularly in children with abdominal tumors, who may have tumor masses that weigh > 10% of their total body weight.17 In such patients, the use of arm anthropometry in assessing nutritional status is valuable because it is independent of tumor mass.18–20

Assessment of nutritional status is difficult because there is currently no gold standard.21 It is multidimensional and can be evaluated by dietary, clinical/ anthropometric, biochemical, and other methods. The usefulness and limitations of dietary assessments have been reviewed.22 To our knowledge to date, there is still little information on formal comparisons of different measures23 that could define the best way to determine nutritional status. In fact, there is some evidence that confidence in simple measures of nutritional status is misplaced when better methods of assessment are used. In a large prospective, controlled study17 of 100 newly diagnosed children with cancer, the mean WFH ratio did not differ from the control reference values. However, the results of arm anthropometry differed markedly between the patients and control values: Twenty-three percent of patients had triceps skinfold thickness (TSFT)—a measure of fat mass—more than 2 SD below the mean value for controls, and 20% had middle-upper arm circumference (MUAC)—a measure of lean body mass—less than the 5th percentile of the control distribution. In another study,19 the result was very similar; whereas no statistically significant differences were found in the WFH values between the patients and the control group, TSFT and MUAC measurements were significantly lower among the children with cancer (P < 0.001), and this was even more evident in patients with intraabdominal solid tumors. A third study20 found that, in a cohort of 40 children, 23% were malnourished by WFH criteria, and 32% were malnourished according to arm anthropometry.

Biochemical tests have limited usefulness in the determination of nutritional status,24 although claims to the contrary (for prealbumin) have been made for both ALL25 and solid tumors26 in children. In particular, although plasma proteins, such as albumin, retinol-binding protein, transferrin, and prealbumin, represent visceral protein, these are also acute-phase reactants. Therefore, their levels may be altered by other factors, such as fever and infection.27 In children with cancer, the concentrations often are depressed, but they do not correlate with other indices of nutritional status.28 Nevertheless, there may be a role for such biochemical measurements in the serial assessment of nutritional status in children with cancer.

Other methods of assessing nutritional status include measurement of body composition by determination of total body water (by deuterium or 18O dilution), bioelectrical impedance, neutron activation, estimation of 40K, and dual-energy X-ray absorptiometry (DXA).24, 27 DXA scans are fast becoming accepted standards in clinical practice,29 with a precision of < 2%.30 This technique has been used to assess the body composition of children with ALL both during31 and after29 therapy.

In summary, these is a clear need to generate normative data for growth and body composition, especially for the populations of children (in developing countries) that contribute the great majority of cases of cancer. True body weight (and, thus, also BMI) may be confounded by large tumors, whereas arm anthropometry offers more sensitive assessments of nutritional status that are not so confounded. MUAC and TSFT measurements can be performed accurately and reliably, even in developing country settings, as exemplified in our experience with colleagues in Latin America (notably, in Cuba and Uruguay). The place of biochemical measures of nutritional status remains to be defined fully, although, in our recent studies, we have discovered that serum creatinine is a good surrogate for lean body mass measurement and may be particularly useful when DXA scans are not available for this purpose.

Pathogenesis of Malnutrition

The pathogenesis of the energy imbalance that underlies the development of malnutrition in any disease, including cancer, is illustrated in Figure 1. This imbalance is the result of some combination of diminished intake, enhanced losses (including malabsorption), and increased needs. Many patients with cancer suffer anorexia and, thus, have reduced intake; others experience increased losses, and some have increased energy expenditure.

Figure 1.

The pathogenesis of energy imbalance in chronic disease. Reprinted from Wilson DC, Pencharz PB. Nutritional care of the chronically ill. In: Tsang RC, Zlotkin SH, Nichols BL, Hansen JW, editors. Nutrition during infancy: principle and practice, 2nd ed. Cincinnati: Digital Educational Publishing Inc., 1997:38, with permission from Elsevier, Inc.

Changes in the metabolism of fat, carbohydrate, and protein have been demonstrated in the cancer-bearing host.32 These changes include increased lipid breakdown, resulting in depletion of lipid stores,33 and alterations in carbohydrate metabolism, resulting in an energy-losing cycle.34 In addition, there is an increased protein turnover35 and loss of the normal compensatory mechanism seen in starvation. The final result is weight loss, in particular, loss of lean body mass, which is manifest clinically as malnutrition (i.e., PEM).

Some patients with cancer have a higher caloric expenditure than patients without cancer.36 In fact, cancer patients use both dietary glucose as well as glucose produced by gluconeogenesis and from amino acids.37 Glucose is transformed into lactate by the tumor. The lactate must then be recycled by the liver, at a large energy cost. This process, the Cori cycle, is increased significantly in patients with advanced cancer.38 It requires the use of muscle proteins, as well as a large proportion of amino acid intake, for gluconeogenesis (Fig. 2). At the same time, there is a severe decrease in total body fat that also may be ascribed to the production of cachectin and tumor necrosis factor by the normal macrophages in response to the tumor.39, 40 Other monocyte mediators, such as interleukin 1 and interleukin 6, can cause metabolic changes, including breakdown of protein and decreased synthesis of albumin.32, 41

Figure 2.

Changes in the metabolism of fat, glucose, and amino acids (AA) induced by the presence of the tumor. ADP, adenosine diphosphate; ATP, adenosine triphosphate; FFA, free fatty acids; TNF, tumor necrosis factor. Reprinted from Yu LC. Nutrition and childhood malignancies. In: Suskind RM, Lewinter-Suskind L, editors. Textbook of pediatric nutrition, 2nd ed. New York: Raven Press, 1993:418, with permission from Lippincott Williams & Wilkins.

The most common risk factors associated with the development of PEM include irradiation to the gastrointestinal tract, intense frequent courses (intervals ≤ 3 weeks) of chemotherapy in the absence of corticosteroids, major abdominal surgery, advanced disease, and lack of a family or health care support system.42 Alterations in taste, anorexia, mucositis, emesis, and diarrhea are other important contributory factors.43 Nutrient deficiency develops over a period of time, depending on the degree of negative balance and the amount of energy reserve available (Fig. 1).

In summary, patients with cancer all too often experience energy imbalance, including increased breakdown of fat and protein as well as energy-consumptive changes in carbohydrate metabolism. These result in net energy loss. The outcome is reduction in weight, particularly lean body mass. Identification of the contribution of various risk factors provides the possibility of designing effective strategies of amelioration for individual patients.

Nutritional Status at Diagnosis

Different studies conducted with different measures have shown that, overall, the prevalence of malnutrition at diagnosis is highly variable, even in developed countries, ranging from < 10% in patients with standard-risk ALL to 50% in patients with advanced neuroblastoma.5, 6, 44, 45 Most detailed information relates to children with ALL. In 1 study,46 105 Mexican patients (53 with high-risk ALL and 52 with low-risk disease) were enrolled. Measurements of MUAC and TSFT were performed monthly for 3 months. At the time of diagnosis, there was no evidence of malnutrition. In another study,47 the nutritional status of 173 consecutive Italian children with newly diagnosed leukemia were assessed by anthropometric measurements, including weight, height, weight for height, MUAC, and TSFT, was compared with the nutritional status of 307 children with nonmalignant diseases. No differences were observed between the two groups. Furthermore, no differences were found when children with high-risk leukemia were compared with children who had standard-risk disease. In a small study48 of 21 Spanish children with cancer, 3 children (14%) showed slight malnutrition on anthropometric evaluation at the time of diagnosis. Finally, anthropometric data, fat-free mass by bioelectrical impedance, energy intake, resting energy expenditure, and biochemical indices were determined at diagnosis in 15 French children with low-risk ALL.49 A group of 15 healthy control children were matched for age and gender. The results showed that there was no indication of undernutrition at diagnosis in the patients as a group.

In summary, these studies suggest that malnutrition is not a prevalent problem in children with ALL at diagnosis. However, there is an obvious need to undertake further detailed studies in children with solid tumors, in whom the expectation of nutritional morbidity at diagnosis is greater, particularly in those with advanced disease.

Nutritional Status during Treatment

Three of the four groups of patients in the studies described above also were evaluated during the first phase of treatment. During the second month after diagnosis, high-risk patients in the Mexican study46 had a greater frequency of altered nutritional status (defined as the loss of ≥ 10% of the arm muscular area) compared with low-risk patients, perhaps due to the more aggressive chemotherapy received by the high-risk group compared with the low-risk group. This interpretation is supported by the results in the French patients49 with low-risk ALL: On Days 22, 36, and 71 of treatment, those children did not develop signs of malnutrition. During the first phase of treatment, the Spanish children48 with modest malnutrition and those with a high risk of becoming malnourished (infants with advanced abdominal disease) were given dietary supplements. In follow-up, 90% of patients showed normal somatic indices. The recovery was most striking in patients with ALL. In summary, treatment of cancer in children may increase nutritional morbidity in the short term, raising the challenge of providing appropriate nutritional intervention (see below).

Nutritional Status and Treatment Tolerance

Malnutrition has been associated causally with the need to reduce the dose intensity of chemotherapy in patients with ALL, including reports on children with standard-risk disease in Mexico50 and those with high-risk disease in Canada.51 In a randomized study of 17 children with Stage IV neuroblastoma, patients who had a favorable nutritional course during the first 21 days of therapy had significantly fewer treatment delays and fewer drug dose reductions throughout the first 10 weeks of treatment.52 Three other randomized prospective studies have documented treatment tolerance benefit from central parenteral nutrition (CPN) compared with an oral, ad libitum diet; improved ability to deliver chemotherapy in 35 children with metastatic disease involving bone,53 accelerated recovery of normal bone marrow in 10 children with acute nonlymphoblastic leukemia,54 and improved adherence to schedules of chemotherapy in 25 children who received abdominal irradiation.55 In a study of 13 patients at high risk for malnutrition who received treatment for Wilms tumor, 7 patients were randomized to receive CPN, and 6 patients were randomized to receive peripheral parenteral nutrition (PPN) plus enteral supplementation (ES).56 Dietary, anthropometric, and biochemical data were determined. CPN was superior to PPN plus ES in reversing PEM (P < 0.05) and prevented delays in chemotherapy and radiotherapy due to granulocytopenia. Twelve low-risk patients received ES only. Although this was ineffective in preventing nutritional depletion and treatment delays in the first 5 weeks, it was effective thereafter. In contrast, another study, which included 27 older teenagers and young adults (median age, 17 years) with poor prognosis sarcoma who received extremely intensive treatment, failed to document benefit from CPN in improving recovery from severe myelosuppression.57 The relative merits of enteral and parenteral nutritional support in critically ill patients have been addressed in a recent editorial,58 and the specific value of adding glutamine to parenteral nutritional support in children has been reviewed.59

It must be recognized as a counterpoint that excessive weight gain may occur during the treatment of ALL, especially in children who receive dexamethasone.60 The mechanism appears to relate to steroid-induced inhibition of growth hormone responses mediated by an enhancement of the somatostatin effect at the pituitary level.61

In summary, there is some evidence that tolerance of chemotherapy is compromised by malnutrition. The ability to reverse or prevent this adversity is related inversely to the intensity of chemotherapy and directly to the assertiveness of the nutritional intervention. Considerable opportunities exist to examine these correlations more exhaustively.

Nutritional Status and Other Morbidity

Compared with adults, children need to continue to grow and thrive during anticancer therapy. In fact, PEM is associated with impaired immune competence, increased susceptibility to infections, and major organ dysfunction.32 In a study of 18 malnourished children with newly diagnosed advanced solid tumors and recurrent leukemia-lymphoma, anergy was documented in 17 patients, and it was reversed with 28 days of CPN support in approximately two-thirds of patients, despite continuing oncologic treatment.62 A close association between an increased risk of infection and malnutrition was demonstrated by a high rate of opportunistic Pneumocystis carinii pneumonia among children with neoplastic diseases.63 Compared with matched controls who had similar tumors, 44 children with cancer who acquired Pneumocystis infection had significantly lower mean weight (P = 0.05) and lower serum albumin levels (P = 0.001). Furthermore, Pneumocystis carinii was found in 3 of 39 South African children who died with kwashiorkor but in none of a comparable group of children who died well nourished. In another study,64 among 20 patients with leukemia, a statistically significant, inverse correlation (P < 0.05) was found between nutritional status and ensuing infection rate. No such correlation was found in 30 patients with solid tumors. A similar experience in children with ALL has been reported by others.51 Another study found that patients who developed cardiomyopathy as a consequence of anthracycline treatment were more likely to have been malnourished at diagnosis or at the initiation of therapy.65

In summary, the interrelation of malnutrition, diminished immunity, and increased risk of infection appears to be well founded. Nutritional supplementation may reduce this risk, but to our knowledge insufficient evidence is available to date. Information concerning other organ dysfunction from malnutrition in children with cancer is even more scarce; thus, there is a pressing need to shed further light on this form of comorbidity and to design studies to prevent or reduce these adverse effects through the agency of nutritional supplementation strategies.

Nutritional Status and Outcome

Nutritional status has a prognostic effect on outcome of therapy in children with cancer. Patients with solid tumors and lymphomas who are malnourished at diagnosis have a poorer survival rate compared with their well nourished counterparts. The relation between nutritional status and outcome is strong in patients with localized solid tumors but less so in patients with metastatic disease. In a retrospective analysis of 244 children with cancer,66 patients were considered malnourished if their weight-height ratio was < 80% of the median weight-height ratio for their age and gender. Nutritional status was related directly to freedom from recurrence among the children with solid tumors overall, whether they had localized or nonlocalized disease. In addition, improved survival was related to good nutritional status in children with localized disease or lymphomas but not in children with advanced disease.

The prognostic effect of nutritional status at diagnosis has been emphasized by a study of 18 children with newly diagnosed Stage IV neuroblastoma.52 Equal numbers of patients were assessed at diagnosis as malnourished or well nourished; significantly more malnourished patients developed recurrent disease or died by 1 year into treatment. Similar results were found in two other studies: The first study was conducted in 167 Brazilian children with ALL.67 A height-for-age Z-score ≤ 1.28 and low socioeconomic status were predictors of recurrence. The second study involved 43 Mexican children with standard-risk ALL68 and found that the 16 undernourished children had a significantly lower rate of 5-year disease free survival compared with the well nourished group (26% vs. 83%). Malnutrition was defined as weight at least 10% below the minimum ideal weight for age and gender minus 1 SD and definite clinical signs of malnourishment. Another case–control study,69 in which 17 Mexican patients with ALL who were receiving remission induction therapy were compared with 76 patients in a control group who had survived at least the phases of induction and consolidation, showed that the chance of dying during the initial phase of the treatment was 2.6 times greater (95% confidence interval, 0.55–11.89) in undernourished children.

However, two studies in other countries with limited resources showed that nutritional status had no demonstrable association with overall survival rates. One study was conducted among children with Wilms tumor in South Africa,70 and the other study was conducted among 443 patients who were treated for cancer (283 patients with leukemia and 160 patients with solid tumors) in San Salvador, in El Salvador, and in Recife, Brazil.71 In the latter study, Z-scores were calculated for weight for age, height for age, and weight for height at the time of diagnosis (Z-scores ≤ − 2 indicated malnutrition). Another study of 1025 children with ALL hat was conducted in the U.K. found no relation between nutritional status based on BMI and prognosis (P = 0.72).72 Those results did not support the earlier findings by the same group of investigators, who interpreted the disparity as related to a small sample size and an extreme of sampling variability.73 Recently, in a publication from Turkey,74 the nutritional status of 47 patients with cancer was evaluated at diagnosis, 3 months after initiation of therapy, and at the end of treatment. Weight for height, height for age, and weight for age were expressed as percentages of standard values. The 3-year overall survival and event free survival rates were determined according to nutritional status. Although the prevalence of malnutrition at the third month was significantly greater than at diagnosis (P = 0.001), the mean Z-scores of the nutritional indices were not significantly different, and the survival rates were not different between malnourished children and well nourished children.

Obesity, as a corollary, is a common finding in survivors of ALL. It is especially common in those who received cranial irradiation, to which it bears a dose-response correlation.75 An important contributory mechanism is likely to be neurosecretory hormonal dysfunction, especially with respect to growth hormone.76 However, an additional factor is reduced physical activity,77, 78 in which treatment-related cardiopulmonary damage may play a role.77–79 Certainly, children who survive ALL have a low rate of participation in sports.80, 81

There is no evidence of unusually high energy intake in children treated for ALL, and their resting energy expenditure is not abnormal.82 Likewise, diet-induced thermogenesis is normal in these survivors.83 Rather, the excess weight gain reflects reduced total energy expenditure,84 indicative of reduced physical activity. The interest in this area of research is such that childhood leukemia has been proposed as a model of preobesity.84

In summary, the intuitively sensible relation between poor nutritional status and a poor prospect for survival does not receive uniform support from published experience. In large measure, this likely reflects considerable variation in the definition of nutritional status, in the nature and extent of the malignant diseases, in the forms of antineoplastic therapy, and in the details of supportive care. By contrast, the problem of obesity in survivors of ALL is well defined and offers ample scope for a therapeutic intervention based on structured physical activity.

Nutritional Status and Other Factors

Malnutrition correlates with other important variables, such as low socioeconomic status, especially in countries with limited resources.85–87 Socioeconomic indicators of poverty, such as poor housing conditions, low per capita income, and domestic energy consumption, associated with difficult access to communication, transportation, and laboratory facilities, were related significantly to a greater risk of recurrence in a univariate analysis of 167 children with ALL.88 In the same group of children, those with malnutrition received approximately 50% of the optimal doses of doxorubicin, mercaptopurine, and methotrexate. Furthermore, a strong correlation was found between poverty and malnutrition in the group with standard-risk disease: It has been claimed that the main problem was malnutrition, which is inseparable from poverty. In a report from Australia, children from the higher social classes had a significantly better 5-year survival rate and duration of first remission compared with children of the lower social classes.89 There are at least two possible mechanisms to explain these findings: physicians' failure to adhere to the doses and schedule required by the treatment protocol90 and/or the child's or family's noncompliance with therapy, in particular during the maintenance phase. Noncompliance with treatment is a strong hypothesis to explain the marked influence of nutritional and socioeconomic factors on the prognosis of ALL in children, not only in developing countries but also in industrialized societies.91, 92 Again, ethnicity is a potential confounding factor: Asian (Indian and Pakistani) children with ALL living in the U.K. had a poorer prognosis compared with native white children93 with ALL, and there are differences in the relative distribution of biologic subtypes in children from diverse socioeconomic strata.

Other factors to consider that may influence the correlation between nutrition and cancer in children are advanced diseases at diagnosis, variability in drug metabolism,94 and a high rate of abandonment of treatment.95 In summary, aside from biologic factors, the influence of nutritional status on the course of cancer in children may reflect a broader socioeconomic disadvantage and be mediated by a phenomenon (reduced compliance or noncompliance with therapy) that is a proper target for remediation.

DISCUSSION

So, what do we know and what do we need to know about this dynamic triangle of children, cancer, and nutrition? The manifest inconsistencies among published reports on this topic reflect 1) different definitions of nutritional status, 2) the wide variety of measures used for assessment, 3) disparate rates of nutritional morbidities in the general childhood populations from which the study samples are drawn, 4) selection biases by disease and stage, 5) small sample sizes, and 6) treatment protocols of variable dose intensity and efficacy.

However, supportable deductions from the available evidence include the following: 1) nutritional status must be defined with respect to local norms; 2) arm anthropometry offers advantages over measures of height and weight and provides useful assessment of nutritional status, especially in developing country settings; 3) the greater the tumor burden and more dose-intense the therapy, the higher the risk of nutritional morbidity; and 4) dietary supplementation, particularly by the parenteral route, can reverse malnutrition and improve the tolerance of chemotherapy. So, why are there no reports on nutritional intervention improving survival rates in children with cancer?

Malnutrition is only one element among socioeconomic disadvantages that are associated negatively with many components of cancer control, from access to care,96 through treatment compliance, to long-term follow-up. Compliance of health care providers and consumers alike is a promising target for investigation. Investigators at St. Jude Children's Research Hospital reported a significant improvement in the survival of black children with ALL in successive eras of therapy97 without any change in the proportion of patients who exhibited subnormal nutritional status, defined on the basis of weight for height.98 This accomplishment was not matched by contemporaneous national experience in the U.S.98 and may reflect the laudable admission and treatment policies of the institution in Memphis.

The recent establishment of a focus on nutrition by the Children's Oncology Group offers the prospect of resolving this conundrum and others, to the ultimate benefit of children with cancer and their families. On balance, it appears likely that the coexistence of diminished nutritional status adds to the risks of morbidity and mortality in children with cancer due to the diseases and their treatment. These risks occur both in the short term, as in the increased rate of comorbidities such as infection, and in the form of late sequelae, such as cardiomyopathy. Acceptance of those added burdens resulting from malnutrition stimulates the desire for effective nutritional interventions. Like other areas of clinical practice, the absence of level one evidence demands the coalescence of opinion to form guidelines of management by consensus. A prime example of the success of such a process is the development of a nutritional algorithm by the Metabolic and Infusion Support Service at St. Jude Children's Research Hospital.99 This provided a rationale for intervention and challenged the staff to overcome inconsistencies in practice. A major outcome was that the relative use of ES (compared with use of the parenteral route) increased three-fold. In fact, because of the complications associated with parenteral nutrition (CPN/PPN), ES delivered by nasogastric tube or gastrostomy tube feeding always should be considered first. Some advantages of ES compared with CPN/PPN include better maintenance of the structural and functional integrity of the gastrointestinal tract, a decreased risk of bacterial translocation, greater ease and safety of administration, and improved cost-effectiveness.100 Consequently, CPN/PPN should be consider only if ES is not sufficient for nutritional needs.

Without wishing to be proscriptive, we suggest that there is an obvious framework for a research agenda in this area. Adjunctive studies of nutritional status and nutritional interventions should be undertaken in the context of large, randomized clinical trials for the treatment of cancer in children, especially those with solid tumors. By design, these will involve well defined clinical groups and antineoplastic strategies. Longitudinal studies will be particularly informative and may yield valuable information even beyond the period of active therapy, in the exponentially growing population of long-term survivors.

Outcome measures, in addition to survival, could include comorbidities, such as end-organ (e.g., cardiac) dysfunction, and health-related quality of life as well as changes in nutritional status. It is likely that major foci of interest will be the sensitivity and specificity of measures of nutritional status and the comparison of nutritional supplementation by enteral and parenteral routes. Economic evaluations, in the form of cost-effectiveness and cost-utility analyses, would be logical undertakings. These studies need to be matched by the generation of normative data on growth and body composition, notably in children living in countries with limited resources where the biggest burden of cancer in childhood is borne and where malnutrition and its correlated deprivations are most prevalent.

Ancillary