Body Composition, Metabolic Health, and Functional Impairment among Adults Treated for Abdominal and Pelvic Tumors during Childhood

This article is featured in Highlights of This Issue, p. 1697

Improvements in treatment have resulted in 5-year survival rates of more than 80% for children diagnosed with cancer. Unfortunately, these survivors are at increased risk for developing abnormalities in body composition, including obesity, dyslipidemias, and insulin resistance (IR). While body composition abnormalities and cardiometabolic impairments have been characterized among survivors of pediatric lymphoblastic leukemia (1–3), brain tumors (4), and hematopoietic stem cell transplant (HSCT; refs. 5–7), less evidence is available for survivors of abdominal and pelvic tumors. Body composition abnormalities and cardiometabolic impairments are of particular concern among survivors, given that in the general population these conditions increase the risk of developing life-threatening diseases including atherosclerosis, coronary artery disease, myocardial infarction, stroke (8), and type 2 diabetes (9), and because many survivors have received cancer treatments in childhood that adversely affect endocrine (10) and cardiovascular health (11).

Data from the Childhood Cancer Survivor Study (CCSS) indicate that survivors of Wilms tumor and male survivors of neuroblastoma are more likely to be underweight when compared with siblings when self-reported height and weight are used to determine body mass index (BMI; ref. 12), a measure that does not distinguish between lean and fat mass (13). In addition, studies in childhood cancer survivor cohorts of mixed diagnoses have documented an association between radiation to the abdomen (14) or pancreatic tail (15) and diabetes, and between radiation to the abdomen and dyslipidemia (16) independent of BMI.

Importantly, many previous studies relied primarily on self-report to document body composition and cardiometabolic risk factors among solid tumor survivors, did not include measures of lean mass, fat mass, or laboratory values of cardiometabolic health, or clinically assessed physical performance. To address these deficits, the aims of this analysis among survivors of childhood abdominal and pelvic tumors were three-fold. First, we characterized body composition using dual X-ray absorptiometry (DXA) and identified lifestyle- and treatment-related factors associated with changes in relative lean mass and fat mass. Second, we evaluated associations between body composition and lifestyle on cardiometabolic health. Third, we evaluated whether changes in body composition influenced physical performance among survivors of abdominal and pelvic solid tumors. We hypothesized that abdominal/pelvic solid tumor survivors have decrements in lean mass, but not fat mass, despite having normal, or slightly lower, BMI than expected, and that these changes are associated with exposure to abdominal/pelvic radiotherapy and poor lifestyle habits. Furthermore, we examined associations between body composition with cardiometabolic markers and physical performance to explore the clinical significance of changes in body composition among solid tumor survivors.

Eligibility criteria for this study included individuals diagnosed with a pediatric abdominal or pelvic solid tumor who were previously treated at St. Jude Children's Research Hospital (SJCRH, Memphis, TN), and who were aged ≥18 years and ≥10 years from diagnosis (17, 18). All eligible individuals were members of the St. Jude Lifetime (SJLIFE) cohort who attended SJCRH campus (Memphis, TN) for clinical evaluation prior to June 30, 2014. Diagnoses considered in these analyses included: neuroblastoma; Wilms tumor; hepatoblastoma; germ cell tumor; carcinoma of an abdominal or pelvic organ; osteosarcoma or Ewing sarcoma involving the eight rib and below, lumbar spine, pelvis, or hip; and soft-tissue sarcoma of the diaphragm or abdominal wall. Survivors who underwent an amputation (ICD9 84.01–84.19) were not considered eligible. Study procedures and documents were approved by the institutional review board. Informed written consent for participation in the SJLIFE study was obtained from all participants.

Whole-body DXA performed using a Hologic Model QDR 4500 fan-array scanner (19–22) was used to assess relative lean mass [lean mass (kg) divided by height in meters squared], and relative fat mass [fat mass (kg) divided by height in meters squared]; Z-scores were calculated using sex- and race-specific values from the National Health and Nutrition Examination Survey (NHANES; ref. 23). Specifically, measures from each subject were matched to the corresponding age, sex and, ethnicity from NHANES. BMI was calculated as weight (kg) divided by height in meters squared. Waist-to-height ratio (WHtR) was calculated by dividing the waist circumference (cm), measured at the point midway between the xiphoid process of the sternum and the umbilicus, by height (cm).

Insulin sensitivity was calculated using the Homeostatic Model Assessment (HOMA IR) index formula, glucose (mg/dL) × insulin (mU/L)/405 (24). Participants with a HOMA IR >2.86 were considered to have IR. Diabetes mellitus was defined by presence of one of the following: fasting blood glucose level ≥126 mg/dL on two separate tests, hemoglobin A1C ≥6.5%, random glucose ≥200 mg/dL, or by the use of glucose lowering medications. High low-density lipoprotein (LDL) cholesterol and high triglycerides were defined by a fasting LDL cholesterol ≥130 mg/dL and triglycerides ≥150 mg/dL, respectively, or by the use of cholesterol lowering medications. Low high-density lipoprotein (HDL) was defined as HDL <40 mg/dL among males or <50 among females (25).

Isokinetic knee extension [Newton-meters (Nm)/kg at 60 and 300°/second] and ankle dorsiflexion (Nm/kg at 60 and 90°/second) were performed to measure muscular strength and endurance (Biodex System III; ref. 26). Low back and hamstring flexibility were assessed using a sit and reach test (Flex-tester, Novel Products; ref. 27) and ankle flexibility was assessed using a goniometer (active and passive dorsiflexion and plantarflexion; refs. 28–30). Hand grip strength (kg) was measured using a hand-held dynamometer (Jamar, Patterson Medical; ref. 31). For the 6-minute walk test (6MWT), participants were asked to walk as far as possible on a premeasured walkway (32).

Diagnosis and treatment data were abstracted from medical records. For patients who received radiotherapy (within 5 years of primary childhood tumor), MTD to the abdomen and pelvis was taken as the total prescribed dose from all overlapping abdominal or pelvic radiotherapy fields, respectively. In addition, each participants' radiotherapy fields were reconstructed on age-specific phantoms and mean dose to the tail of the pancreas was estimated using previously described methods (33, 34). Cumulative drug doses for alkylating agents and anthracyclines were converted to cyclophosphamide and doxorubicin equivalent doses (35). Data on lifestyle habits were collected using a structured questionnaire. Smoking status was classified as current, past, or never. Risky drinking was defined as >3 drinks per day or 7 drinks per week for women, and >4 drinks per day, or >14 drinks per week for men. Survivors who met the Centers for Disease Control and Prevention (CDC) guidelines for physical activity (150 minutes of moderate intensity physical activity or 75 minutes of vigorous activity per week) were defined as active (36).

Mean values for body composition outcomes among survivors, stratified by sex, were compared with data from the 2013 to 2014 NHANES using one-sample t tests and the prevalence of IR, low HDL, and high LDL and triglycerides among survivors was compared with NHANES using indirect standardization (37). General linear regression was used to test associations between lifestyle- and treatment-related factors (cumulative drug dose and increasing radiation exposure in 10 Gy increments) with relative lean and fat mass Z-scores among survivors. To facilitate these analyses, we imputed missing values for 74 survivors who did not have DXA measurements using the multiple imputation procedure with fully conditional specification (38, 39) option in SAS 9.4 (SAS Institute). Missing values were imputed using sex and body fat assessed by skin-fold based on findings of a previous study (40). Twenty independent datasets were created, on which regression analyses were run; parameters were summarized by the MIANLYZE procedure, which accounts for within and between imputation variability. Log-binomial regression was used to assess the association between body composition and lifestyle on the RR of cardiometabolic impairments. As relative lean and fat mass are correlated, we calculated a composite variable in which survivors were divided into four groups based on their lean mass and fat mass Z-scores (Fig. 1: body composition analytic groups): high to normal muscle mass and low to normal adiposity (HM-LA, lean mass Z-score > −1 and fat mass Z-score < 1), high to normal muscle and high adiposity (HM-HA, lean mass Z-score > −1 and fat mass Z score ≥ 1), low muscle mass and low to normal adiposity (LM-LA, lean mass Z-score ≤ −1 and fat mass Z-score < 1), and low muscle mass and high adiposity (LM-HA, lean mass Z-score ≤ −1 and fat mass Z score ≥ 1). However, there were only two individuals with lean mass Z-score ≤ −1 and fat mass Z-score ≥ 1, thus, the LM-HA category was dropped from analyses. Finally, potential associations between body composition and mobility and function were assessed using generalized linear models with analyses stratified by sex. All analyses were conducted in SAS 9.4.

There were 727 survivors who met the eligibility criteria for this study, of whom, 431 underwent clinical evaluation and had data available for these analyses (Supplementary Fig. S1). The median age at diagnosis was 3.6 (range, 0–24.8) years, the median age at assessment was 29.9 (range, 18.7–55.1 years). As seen in Table 1, the most frequent childhood diagnoses among participants were Wilms tumor (42.9%), neuroblastoma (16.5%), and germ cell tumor (14.8%). A lower frequency of participants were male when compared with nonparticipants (44.1% vs. 54.0%; P = 0.008). Although, a higher frequency of participants had received radiotherapy when compared with nonparticipants (50.1% vs. 38.5%; P = 0.002), there were no differences in mean abdominal, pelvic, or pancreatic (tail) radiation doses between participants and nonparticipants (P > 0.05).

Table 2 summarizes measures of body composition among solid tumor survivors. Mean relative lean mass was reduced in survivors compared with normative data for both males [−0.67 (±1.29); P < 0.001] and females [−0.72 (±1.38); P < 0.001]. Mean relative fat mass was also lower among survivors than normative data [males, −0.32 (±1.28); P < 0.001 and females, −0.16 (±1.09); P = 0.025]. The prevalence in obese BMI (BMI ≥ 30 kg/m²) was lower among survivors (27.2% vs. 34.5%; P < 0.01) when compared with normative data, while the prevalence of underweight BMI (BMI <18.5 kg/m²) was higher (6.0% vs. 1.5%; P < 0.01).

Supplementary Table S1 (flow diagram of participation) shows differences in body composition among survivors stratified by radiation exposure. Mean lean mass Z-score [−1.15 (±1.41) vs. −0.31 (±1.16); P < 0.001] and mean WHtR [0.47 (±0.09) vs. 0.52 (±0.10); P = 0.012] were lower in survivors treated with abdominal/pelvic irradiation than those not treated with radiotherapy.

Increasing abdominal (β = −0.22; SE, 0.07; P = 0.002) and pelvic radiation doses (β = −0.27; SE, 0.08; P < 0.001) were associated with decreasing lean mass among male survivors (Table 3). Similarly, exposure to abdominal (β = −0.30; SE, 0.09; P = 0.001) and pelvic radiotherapy (β = −0.17; SE, 0.08; P = 0.026) were associated with low relative lean mass among females. Males who were physically active had decreased relative fat mass compared with those who were not active (β = −0.42; SE, 0.16; P = 0.031).

The prevalence of IR was 40.2% [95% confidence interval (CI), 35.5–45.0] among survivors of abdominal and pelvic solid tumors, which was increased relative to the prevalence expected from NHANES (33.8%; P = 0.006). IR was most common among survivors diagnosed with tumors of the bone (52.9%) or soft tissue (44.4%: Supplementary Table S2). Survivors with IR had higher BMI (31.7 [SD±7.5] vs. 24.1 [±5.0] kg/m², P < 0.001) and WHtR (0.56 [±0.10] vs. 0.46 [±0.07], P < 0.001) when compared to survivors without IR.

As seen in Table 4, the risk of IR was increased among survivors with HM-HA (RR = 1.86, 95% CI = 1.51–2.28) and decreased among LM-LA (RR = 0.38, 95% CI = 0.23–0.61) compared to survivors with HM-LA Z-scores in multivariable analyses. Although the low frequency of diabetes mellitus (n = 26) in the cohort prevented extensive multivariable analyses of this outcome, radiation to the pancreatic tail was associated with an increased risk of diabetes (RR = 1.66, 95% CI = 1.29–2.14) after adjusting for relative fat mass. No associations between diabetes and pelvic radiation, smoking history, chemotherapy exposure, or physical activity levels were observed in univariate analyses.

The prevalence of low HDL (28.9% vs. 33.5%; P = 0.046) and high triglycerides (18.4% vs. 10.02%; P < 0.001) among survivors of abdominal and pelvic solid tumors were elevated when compared with normative data, while the prevalence of high LDL was not (18.9% vs. 23.83%; P = 0.26). In multivariate analyses, survivors with HM-HA had an increased risk of low HDL (RR, 1.82; 95% CI, 1.32–2.51) when compared with survivors with HM-LA (Table 3). Males had a higher risk of having high LDL (RR, 2.54; 95% CI, 1.64–3.93) and high triglyceride levels (RR, 2.10; 95% CI, 1.35–3.27) than females. The risk of low HDL (RR, 1.66; 95% CI, 1.21–2.28) and high triglyceride (RR, 1.67; 95% CI, 1.07–2.62) levels were increased among smokers.

As shown in Table 5, when compared with survivors with high/normal lean mass and low adiposity, both males and female survivors with high/normal lean mass and high adiposity performed more poorly on measures of quadriceps strength at 60°/second (P < 0.001) and 300°/second (P < 0.001), and reduced performance on the sit and reach test (P < 0.05) and distance covered in the 6-minute walk (P < 0.01). Handgrip strength was reduced among both males (P < 0.001) and females (P < 0.001) with low lean mass and low adiposity compared with survivors with high/normal lean mass and low adiposity.

In this study, which represents one of the largest cohorts of adult survivors of pediatric abdominal and pelvic solid tumors to have undergone comprehensive metabolic, functional, and radiographic assessments, we characterized the impact of radiation on body composition and its secondary consequences on metabolic health and functional performance. We found that survivors, a median of 30 years at follow-up, had low relative lean mass and a higher prevalence of IR, low HDL, and high triglycerides relative to normative values, and that reductions in lean mass were associated with reduced handgrip strength, which is a marker of increased risk for disability and early mortality (41). While we did not observe a difference in adiposity among survivors relative to normative data, survivors with high relative fat mass had an increased risk of developing poor cardiometabolic profiles and were also at risk of demonstrating reduced strength and physical performance.

In this study, adult survivors of pediatric abdominal and pelvic solid tumors had a mean relative lean mass Z-score more than a half SD below the expected population mean, while mean relative fat mass among survivors was only slightly lower than normative data. A prior report from the CCSS suggested that survivors of Wilms tumor and male survivors of neuroblastoma are more likely to be underweight according to BMI when compared with siblings (9, 12). Our data suggest these differences may be attributable to reductions in lean mass rather than reduced adiposity. However, it is unclear why mean relative lean mass was significantly lower among solid tumor survivors compared with normative values. Among patients treated with total body irradiation, normal BMI, but increased percent body fat, has been observed leading investigators to suggest decreased lean mass among this population (5, 7). However, among pediatric transplants populations, increased adiposity and reduced lean mass may occur as a result of multiple factors, including decreased growth hormone secretion (42, 43), thyroid dysfunction and hypogonadism (44, 45) following irradiation to the hypothalamus and pituitary, thyroid, and gonads, and possibly from muscle loss (46). Radiotherapy has been shown to reduce proliferation and survival of human myocytes in vitro and to cause muscle fiber loss in animals (47, 48). Among solid tumor survivors, abdominal- and pelvic-directed radiotherapy may damage postural muscles (49), or subtly impair sex hormone production (50, 51), ultimately affecting muscle mass. It is also likely that poor lifestyle choices impact relative lean mass among survivors, such that children with suboptimal lean mass following cancer treatment (52) may never recover. Although, we attempted to assess the contribution of physical activity on lean mass, our measure of physical activity may not have adequately captured anabolic activities capable of increasing muscle mass.

In this study, the prevalence of IR among survivors of abdominal and pelvic solid tumors was 40%, which was slightly higher than expected from normative data. Among childhood cancer survivors, the risk of IR has been primarily studied in survivors of acute lymphoblastic leukemia and HSCT, with approximately one-half of survivors developing IR at a median of 10 to 26 years from diagnosis (2, 53). IR is of concern, as a high proportion of individuals who are insulin resistant eventually progress to diabetes. We found that high fat mass was associated with a roughly 2-fold increase in IR risk when compared with survivors with higher muscle mass and low adiposity. Mean BMI and WHTR were higher in survivors with IR than without, and were above standard cutoff points for obesity (i.e., BMI ≥30 kg/m²) and central adiposity (i.e., WHTR ≥0.5) for each measure. This contrasts with a study of neuroblastoma and Wilms tumor survivors, in which waist circumference was reported to be altered among survivors treated with abdominal radiation and as a result a poor marker for metabolic impairment (13). We also observed an increased risk of diabetes among survivors treated with pancreatic radiation, which was independent of relative fat mass. This is not unexpected as the pancreatic tail is the main location of B cells of islet, which produce insulin (14, 15). Collectively, ours and others data suggest that survivors may be at risk of developing diabetes through multiple pathways, either from direct damage to the pancreas following radiotherapy, and following IR as a result of alterations in function and secretions of adipocytes and from increased adiposity.

We found the prevalence of low HDL (29%) and high triglycerides (18%) among survivors were increased when compared with normative data, while the prevalence of high LDL was not (19%). High fat mass among cohort members was associated with low HDL. Obesity, particularly abdominal obesity, has a direct association with dyslipidemias and all these conditions are driven by excessive caloric intake, consumption of foods high in saturated fats, cholesterol, and carbohydrates, and insufficient physical activity. Our data suggest that factors commonly associated with dyslipidemias in noncancer populations also increase the risk of these conditions among survivors of pediatric solid tumors. Nevertheless, prevention and amelioration of these conditions are important among survivors not only because dyslipidemias are associated with increased risk of cardiovascular disease, but also because many solid tumor survivors receive cancer treatments that increase their risk for cardiovascular and renal dysfunction.

A novel aspect of our study is the ability to identify associations between body composition and performance status. High relative fat mass (high adiposity) was associated with deficits in performance across multiple measures of fitness including knee extension strength (at 300° and 60°) and reduced performance on the sit and reach and 6MWTs. High adiposity and associated reductions in strength, mobility, and flexibility among survivors are of concern because these measures are markers of physical fitness; higher levels of fitness are associated with decreased risk of cardiovascular disease (54), hypertension (55), and noninsulin-dependent diabetes mellitus (56) in the general population.

This study has several strengths. First, our population represents one of the largest cohorts of solid tumor survivors to have undergone anthropometric assessment, laboratory testing of metabolic markers, and testing of strength and mobility to date. Second, detailed treatment data, including radiation dosimetry for the abdomen, pelvis, and pancreas, was available for all survivors. However, for comparisons of prevalence, we used data from NHANES. As a result, differences in the measurement of cardiometabolic outcomes among the survivor cohort and NHANES may have adversely influenced comparisons of prevalence. In addition, while we were unable to examine the influence of abdominal or pelvic surgeries on functional performance among survivors, those survivors with serious, severe, or disabling chronic musculoskeletal and neurologic conditions were excluded from analyses.

In summary, we found that survivors of abdominal and pelvic solid tumors had lower relative lean mass than expected and that low lean mass was associated with prior irradiation to the abdomen and pelvis. We also observed a higher prevalence of IR and triglycerides, as well as low HDL among survivors. While increased relative fat mass, and to some extent, low lean mass were associated with poor performance on measures of strength, mobility, and flexibility among survivors, we also observed that high relative fat mass was associated with adverse metabolic profiles including IR. This is important, given that findings from this and other studies indicate that survivors of abdominal/pelvic solid tumors do not have increased BMI, an important risk factors for cardiometabolic diseases, relative to the general population. Moving forward, while it may not be possible to avoid radiotherapy as a key treatment for many solid tumors, further research is required to assess whether interventions in both child- and adulthood could remediate abnormalities in body composition and cardiometabolic impairments. For instance, interventions directed at lifestyle behaviors including adherence to a heart-healthy diet, regular physical activity, maintenance of a healthy weight, and avoidance of tobacco products have been successful in improving lipid parameters, insulin sensitivity, and fitness among non-cancer populations and represent key areas of potential research among pediatric cancer survivors. Ultimately, good lifestyle choices sustained over the long-term may prevent or delay the onset of IR and dyslipidemia among survivors and minimize the risk of future diabetes and cardiovascular disease.

No potential conflicts of interest were disclosed.

The funders had no role in the design of the study; the collection, analysis, or interpretation of the data; the writing of the communication; or the decision to submit the communication for publication.

Conception and design: C.L. Wilson, W. Chemaitilly, D.K. Srivastava, M.M. Hudson, L.L. Robison, K.K. Ness

Development of methodology: C.L. Wilson, W. Chemaitilly, M.M. Hudson, K.K. Ness

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M.M. Hudson, L.L. Robison, K.K. Ness

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C.L. Wilson, W. Liu, W. Chemaitilly, C.R. Howell, D.K. Srivastava, R.M. Howell, M.M. Hudson, K.K. Ness

Writing, review, and/or revision of the manuscript: C.L. Wilson, W. Liu, W. Chemaitilly, C.R. Howell, D.K. Srivastava, R.M. Howell, M.M. Hudson, L.L. Robison, K.K. Ness

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C.L. Wilson, L.L. Robison

Study supervision: C.L. Wilson

Other (team did the radiation dosimetry for the CCSS cohort, reviewed/edited the manuscript): R.M. Howell

This project was funded by Cancer Center Support grant number CA021765 (to principal investigator, C. Roberts), CA195547 (to multiple principal investigators, M.M. Hudson and L.L. Robison), and the American Lebanese Syrian Associated Charities.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.