Main

Growth failure and malnutrition are common clinical features in CF(1–5). These features are associated with pulmonary morbidity and mortality(6–8), but the nature of this association is poorly understood. According to the 1994 CF Foundation Patient Registry(9), 29% of patients are in less than the 10th percentile for height for age and gender, and 21% are below the 5th percentile. The distribution for weight is similar with 35 and 26% below the 10th and 5th percentiles, respectively(9). In children with CF, growth failure and malnutrition are due to chronic negative energy balance, caused by decreased caloric intake, increased energy loss, increased energy expenditure, or some combination of these factors(10).

Increased REE among subjects with CF has been demonstrated in cross-sectional studies(4, 5, 11–16) and is a contributory factor to growth failure and malnutrition. Increased REE is associated with declining pulmonary function(13), but it is unclear whether increased REE is a cause or consequence. Nutrition intervention studies have suggested that improved nutritional status can slow the progression of pulmonary disease and incrase life expectancy(3, 6, 7, 17–20). However, these studies do not explain the etiology of malnutrition or its role in disease progression in CF.

The purpose of this study was to examine prospectively the relationships among REE, growth, nutritional status, and pulmonary function in a sample of children with CF who, at the time of recruitment, were prepubertal with adequate growth status, and had pancreatic insufficiency and mild pulmonary disease. Baseline comparisons with healthy control subjects(11, 21, 22), showed that both groups were similar in most, but not all, measures of nutritional status. The CF and control groups were similar in height, weight, and FFM, but differed in percent body fat (CF, 13%; controls, 17%) and triceps Z score (CF,-0.4; controls, 0.0). REE was significantly greater (6-9%) in the CF group. This difference was not associated with the ΔF508 homozygous genotype.

The changes that occurred in these CF and control children in REE, nutritional status and pulmonary function over the subsequent 3 y are reported here. The study was designed to determine whether REE increased over time and if this increase was associated with significant changes in nutritional status and/or pulmonary function. In addition, this study examined the nature of the relationship between changes in REE, growth, nutritional status, and pulmonary function to understand better the etiology of malnutrition in CF.

METHODS

Subjects

The subjects with CF were recruited from the CF Center at the Children's Hospital of Philadelphia in 1988-1990. CF was diagnosed based on clinical signs and duplicate quantitative pilocarpine iontophoresis sweat tests with sodium and chloride values greater than 60 mEq·L-1. The patients had to meet the following criteria at entry: ages 5-10 y, pancreatic insufficiency, prepubertal, FEV1 > 60%, no acute pulmonary exacerbations or other illnesses at the time of the study, growth in height and weight above the 3rd percentile, and no other chronic disorders or prescribed medication known to affect growth or nutritional status. A group of 26 healthy children of comparable age, gender, and weight to the children with CF were recruited as control subjects from hospital and community sources. Subjects were evaluated four times at yearly intervals while in their usual state of good health.

Fifty-two children receiving care at the CF Center were in the appropriate age range. Of these, eight did not meet other eligibility criteria and 19 families declined to participate. Twenty-five children (12 boys and 13 girls) with CF met all the eligibility criteria and were recruited into the study. Fourteen children had the ΔF508/ΔF508 genotype, seven had theΔF508/other, and four the other/other genotype. Male and female patients were evenly distributed in the genotype groups. The control group consisted of 26 healthy children (15 boys and 11 girls) of similar age, gender, and racial distribution. There was one child with African ancestry in each group, and all others were of European ancestry. All children were prepubertal at the time of recruitment. The sample size was sufficient to detect a difference between CF and control groups in REE at baseline of 8 ± 10% relative to control subjects with 80% power while controlling for type I error of 0.05.

The protocol was approved by the Institutional Review Board at the Children's Hospital of Philadelphia. Informed written consent was obtained from parents and assent from the children. The studies were conducted in the Nutrition and Growth Laboratory and the General Clinical Research Center of the Children's Hospital of Philadelphia during an 18-h overnight admission.

Anthropometry and Tanner score. The anthropometric evaluation consisted of measurement of weight, height, midarm circumference, and triceps skin fold thickness in triplicate using standard methods as previously described(21, 23). All growth measurements were made by one well trained anthropometrist in the first year, and by a second well trained anthropometrist in subsequent years. Tanner stage of pubertal development was assessed by report. A score was assigned representing an average of genital and pubic hair development.

Body composition by TBW. TBW was measured by the deuterium oxide (2H2O) dilution method(24). Each subject received an oral dose (0.14 g of 2H2O·kg estimated TBW-1)(25) of 99.8 or 99.9 atom percent deuterium (Aldrich, Milwaukee, WI). A baseline urine sample was obtained, and then deuterium oxide was administered orally 1.5 h after a standardized evening meal. Overnight urine and the first urine void of the morning(26) were collected after an overnight fast and pooled. The volume and concentration of the pooled sample were used to adjust for isotope losses during the equilibration period. The second urine void of the morning was collected and analyzed to determine TBW. Urinary2 H2O isotopic abundance was measured by isotope ratio mass spectroscopy(11) (Nuclide 6-60, State College, PA) by Global Geochemistry Corporation (Canoga Park, CA) at baseline, and by Metabolic Solutions (Merrimack, NH) in the subsequent years of the study. Duplicate samples were analyzed by each company to assure agreement between laboratories. A hydration factor of 75.3% for prepubertal children was used to determine FFM, fat mass, and percent body fat from TBW(27).

Pulmonary function status. Pulmonary disease was evaluated at the time of the study in all subjects with CF using standard pulmonary function methods for spirometry(28, 29). FVC, FEV1, and FEF25-75 were compared with appropriate reference values and expressed as a percent of the reference values.

Resting energy expenditure. REE was measured by indirect calorimetry (Gould 2900 or SensorMedics 2900Z, SensorMedics, Yorba Linda, CA) as previously described(11). Each child was given a standardized evening meal and fasted from food and medication for 12 h before the REE measurement. In the early morning the child was awakened, a normal axillary temperature was documented, and physical activity was restricted to bedrest. After an initial 10-min adjustment period, data were collected for 50 min while the child was awake and resting quietly watching a videotape. Data points derived during periods of significant physical movement or coughing were eliminated, and the remaining data points were averaged. The coefficient of variation for the REE measurement itself was 5 ± 3% based on REE measurements recorded under similar conditions in seven children with CF (ages 9-15 y) measured on three consecutive days during a clinically stable period.

Genotype analysis. Genomic DNA was isolated from blood samples(Applied Biosystems, model 340A Nucleic Acid Extractor, Foster City, CA) and analyzed as previously described(10). The presence of the ΔF508 mutation was determined by allele-specific oligonucleotide hybridization as described in Kerem et al.(30). Subjects with CF were classified as ΔF508 homozygotes or other CF genotypes, including ΔF508 heterozygous, and other non-ΔF508 mutations. Control subjects were not evaluated.

Reference Values

National Center for Health Statistics reference standards(31) were used to calculate Z scores(standardized scores) for the anthropometric data. For height and weight, these were calculated using the Anthropometric Software Program, Version 3.1, provided by the Centers for Disease Control (CDC, 1988, Atlanta, GA). IBW was determined as the weight that corresponded to the weight for age percentile equal to the actual height for age percentile on the National Center for Health Statistics growth chart(32). Percent IBW was calculated as the observed weight divided by the IBW and multiplied by 100. The triceps Z score was computed using reference data from Frisancho(33). REE was expressed both as kilocalories/d and as percent of the predicted values from World Health Organization equations(%pred WHO)(34).

Statistical Analysis

Confidence intervals of measures of growth, body composition, nutritional status, and pulmonary function by group and gender at each time point were calculated and plotted to examine time trends descriptively. The presence of longitudinal trends was analyzed based on the longitudinal mixed effects approach of Laird and Ware(35). This model extends the usual repeated measures approach in analysis of variance by enabling us to associate linearly the repeated outcome variable with a continuous explanatory variable, such as time of measurements, and to include in the analysis all observed subjects, even when some subjects do not have a complete set of measurements. This approach has the additional advantage of weighing each subject proportionally to his/her contribution to the model based on the number of observations and the precision of the observations. In other words, if a subject's observations are extremely variable, the slope of this subject is weighted less than that of a subject with less variable observations. Each subject has a random effect in the model, in addition to the usual fixed effects. Calculations were based on Laird et al.(36) and performed in BMDP5V statistical package(37). Models were fit to growth, energy expenditure, and pulmonary function to test for change over time, and whether patterns of change were different between the CF and control groups and between male and female subjects. Time was measured as time since study entry. Models explaining prospective relationships between changes in energy expenditure and changes in other parameters (e.g., growth, nutritional status) were fit, as well as models for changes in pulmonary function. All models were stratified on group and gender and allowed for interaction terms. The effect of genotype on REE and pulmonary function was evaluated also.

RESULTS

Subject characteristics. Over the 3 y of the study, a total of 195 evaluations were completed. The final sample was restricted to those observations for which both REE and TBW measurements were available, resulting in a total of 176 observations. In the final year of the study, six girls with CF had started puberty (four in Tanner stage 2; two in Tanner stage 3), and five control children (two boys and three girls) had started puberty (Tanner stage 2).

Tables 1 and 2 show the mean (± S.D.) values for growth, nutritional status, REE, and pulmonary function for the children with CF and controls at each visit. Figure 1 shows the mean and 95% confidence intervals for height and weight Z scores and REE (expressed as%pred WHO). The CF and control groups were similar in weight and weight Z scores. Height Z scores were consistently lower in the CF boys than in the control boys throughout the study. REE was, on average, higher for both boys and girls with CF than the control group, although there was considerable overlap. Boys with CF also had higher REE (expressed as%pred WHO), lower percent body fat, and tricepsZ score than the control boys.

Table 1 Longitudinal growth, nutritional status, body composition, and pulmonary function characteristics (mean ± SD) of children with CF and healthy controls
Full size table
Table 2 Prospective changes in pulmonary function(percent predicted) for children with CF
Full size table
Figure 1
figure 1

Changes in height Z score, weight Z score, and resting energy expenditure as percent of predicted using World Health Organization prediction equations (mean and 95% confidence interval) over 3 y for children with cystic fibrosis (•―•) and healthy controls(○- - - -○).

Full size image

Longitudinal changes in growth, nutritional status, energy expenditure, and pulmonary function parameters. Longitudinal mixed effects models combining the data from all four measurements showed that weightZ score declined significantly in boys with CF (p = 0.002) at a rate of -0.09 Z score units per year. Control boys experienced a numerical (-0.04 Z score per year, p = 0.06) decline in weight Z score. Girls with CF maintained their weight status, but the control girls had a significant (p = 0.001) decline in weightZ score of -0.09 Z score units per year.

Height Z scores revealed different patterns. The boys with CF did not change in height Z score. The control boys increased in heightZ score by 0.05 Z scores per year (p = 0.007). Girls with CF experienced a decline in height Z score of -0.11Z scores per year (p = 0.003), and control girls did not change. Although there were differences in the patterns of height and weightZ scores between CF boys and girls over time, both boys and girls with CF had similar growth status, as indicated by height and weightZ scores, at the end of the study (-0.50 to -0.60 Z scores).

Triceps Z score and%IBW followed longitudinal patterns similar to weight Z score. Triceps Z score decreased significantly(p < 0.00005) and to a similar degree in both CF and control boys, approximately -0.15 Z score units per year. Girls of both groups maintained their triceps Z score. For CF and control boys,%IBW decreased -1.2 units per year (p = 0.002). Control girls also experienced a significant (p = 0.01) decline in%IBW of -1.43 units per year. Girls with CF maintained their nutritional status throughout the study as indicated by these measures.

REE, expressed as%pred WHO, increased significantly in both the CF(p = 0.004) and control groups (p = 0.03). These changes were due largely to changes in REE among the girls of both groups. For girls with CF, REE, expressed as%pred WHO, increased at a rate of 3.3 units of percent predicted per year (p < 0.00005). The increase for control girls also was significant (p = 0.008) but occurred at half the rate (1.7 units of percent predicted per year). Although boys with CF had significantly greater REE, expressed as%pred WHO, than the control boys throughout the study (p = 0.002), neither group changed in REE(as%pred WHO) over time.

No significant change in pulmonary function, as measured by FEV1 or FEF25-75, occurred in the CF group as a whole. FVC declined significantly (p = 0.03) for the group, but this was due to losses in FVC in boys (p = 0.01) and not girls. Similarly, FEV1 did not decline significantly in girls, but it did so for boys (p = 0.03); FEV1 was 94% predicted at baseline and 85% predicted at y 3. FEF25-75 decreased numerically, but not significantly (p = 0.08) in boys.

There were no effects of genotype observed in relation to longitudinal changes in growth, REE, expressed as%pred WHO, or pulmonary function in all the analyses described above.

Prospective relationships between resting energy expenditure, growth, nutritional status, and pulmonary function in children with CF. Several longitudinal models were tested to determine which of the growth and nutritional status variables were the best predictors of REE and FEV1 over time. Analyses of pulmonary function measures were restricted to children with CF. For all other longitudinal analyses, the data for the CF and control groups were combined, and indicator terms were used to test for group (CFversus control) and gender (male versus female) differences. Interaction terms also were included in the models.

Height, FFM, percent body fat, triceps Z score,%IBW, FEV1, and genotype were the explanatory factors initially evaluated in the REE(kcal/d) models. Triceps Z score and FEV1 were not significant predictors of REE and were excluded from subsequent analyses. The final set of models for REE are illustrated in Figure 2. The significance levels and log-likelihood values are shown inTable 3. Percent IBW (model 1) was a significant predictor of REE along with the group and gender indicators. However, this model was not as strong as the other two models, as indicated by log-likelihood values. Furthermore, as illustrated in Figure 2, this model does not account for the observed increases in REE that were expected and occurred in both groups as the children aged and grew from the beginning to the final year of the study (Fig. 2A).

Figure 2
figure 2

Longitudinal mixed effects models (B-D) predicting changes in resting energy expenditure (kcal/d) for children with cystic fibrosis (boys, ▪―▪; girls•―•); and healthy control subjects (boys, □- - - -□; girls, ○- - - -○) compared with changes in measured resting energy expenditure (A).

Full size image
Table 3 Longitudinal mixed effects models for resting energy expenditure and pulmonary function in children with CF and healthy control subjects
Full size table

Growth in height was a significant predictor of changes in REE (model 2) and accounted for the increases in REE through the course of the study along with group and gender differences. FFM (model 3) and height (model 2) were comparable predictors of increasing REE over time as indicated by the log-likelihood values. These models were also tested replacing time with age and gave similar results. Model 2 was also analyzed using the baseline values as covariates and three repeated measures. The terms that previously were significant in Table 3 remained significant, and the sizes of the effects were comparable. For every 1-kg increase in FFM, REE increased by 26 kcal/d. Group (CF versus control) and gender differences also were significant such that REE was 106 kcal/d greater in the CF group than the control children given the same FFM. For boys, REE was 65 kcal/d greater than for girls with the same FFM. Interaction terms for group·FFM and gender · FFM were also significant, indicating that, as FFM increased, the differences in REE between groups and genders increased further. Figure 2 illustrates these interactions for the relationships between REE and FFM in male and female patients with CF and control subjects.

Although FEV1 did not decline significantly over time for the CF group overall, longitudinal mixed effects analyses tested whether this measure of pulmonary function was associated with fluctuations in REE or nutritional status in the individual children with CF (see Table 3). REE expressed as%pred WHO was not a significant predictor of changes in FEV1. Change in%IBW was a significant (p = 0.007) predictor of FEV1 over time (model 4). Each unit increase in%IBW was associated with an increase of 0.61 unit of percent predicted FEV1. This relationship was the same for both boys and girls with CF.

The possible effect of genotype was evaluated by testing whether it had a significant effect in the longitudinal mixed effects models described above(see Table 3). There were no significant differences between CF genotype groups in any of the REE models. In the analysis of FEV1, however, genotype was a significant predictor (p = 0.04) along with%IBW (p = 0.005), such that FEV1 was 15 units of percent predicted FEV1 lower in the group homozygous for ΔF508(model 5).

DISCUSSION

Numerous cross-sectional studies have shown that REE is increased by 4-33% in subjects with CF relative to predicted values or control subjects(3–5, 11–16, 38, 39), but its association with clinical status remains controversial. Buchdahl et al.(5) found that REE was related to pulmonary function (FEV1/FVC ratio; r = 0.46, p ≤ 0.05), but not to growth or nutritional status. Bogleet al.(16) also found that REE was elevated but not associated with growth, nutritional status, or measured pulmonary function. However, FVC, FEV1, and FEF50 were significantly associated with weight-for-height (r = 0.43-0.54). Similarly, at baseline in the current study, an association between FEV1 and%IBW was observed(22), but not between REE and pulmonary function(11). In males ages 7-39 y with pancreatic insufficiency, good pulmonary function, and normal nutritional status, Friedet al.(13) found that REE was increased slightly (104-105% pred WHO). In another set of subjects(13) with good nutritional status and advanced lower FEV1 (range 35-70%), REE was greatly increased (125 ± 14%).

The above studies were all cross-sectional in design and varied with respect to sample size, age range, sexual development, nutritional status, and severity of pulmonary disease. Despite their heterogeneity, these studies indicate increased REE in CF, even with mild pulmonary manifestations of CF. The inconsistent associations between REE and measures of pulmonary function are difficult to reconcile, but may be due to variability in nutritional status, age, pubertal status, and gender distribution in the studies. Nutrition intervention studies have shown that it is possible to significantly improve nutritional status and maintain lung function in patients with CF(17–20).

The current study is unique in that a control group and a group of relatively homogeneous prepubertal children with pancreatic insufficiency and mild pulmonary manifestations of CF were examined prospectively. In addition, this study is the first of its kind in using longitudinal mixed effects models to examine interrelationships in the patterns of change in REE, nutritional status, and pulmonary function in children with CF. Girls with CF exhibited an increase in REE expressed as%pred WHO without a statistically significant decrease in pulmonary function. Boys with CF did not exhibit an increase in REE, expressed as%pred WHO, but their pulmonary function declined. At the completion of the study, both boys and girls with CF had similar FEV1 and similarly low growth status in both height and weight.

In this group of children with CF, REE was elevated at baseline relative to controls. Increases in REE and declines in growth status occurred over time without a detectable decline in FEV1 or FEF25-75 for the CF group overall. Changes in REE were not associated statistically with changes in pulmonary function. β-Agonist therapy was examined carefully in relation to REE and FEV1 in the first year of the study, and no association was observed. Given the absence of this initial difference, and the small sample size in this study, this variable was not included in the longitudinal mixed effects models. The models that most closely predicted the observed changes in REE over time included height or FFM, and group and gender as explanatory variables. The differences in REE between CF and control children at baseline became greater as FFM and height increased.

Both REE and pulmonary function were related to nutritional status. Percent IBW and FFM were positively associated with REE, and%IBW was the nutritional status indicator that was significantly (and positively) associated with changes in pulmonary function (FEV1) over time. Children who were not able to maintain their%IBW experienced decreases in pulmonary function.

Although gender differences in life expectancy have been reported repeatedly(40, 41), only the initial report from this study(11) has explored gender differences in REE or disease progression in CF. Prepubertal children were recruited to minimize expected gender effects related to pubertal development. Therefore, gender differences in growth, nutritional status, and REE were unexpected in this sample. Boys with CF declined in growth status and pulmonary function without a significant change in REE expressed as%pred WHO. Girls in both the CF and control groups experienced increases in REE as%pred WHO. The girls with CF declined in growth status but did not exhibit a statistically significant decline in pulmonary function. This gender difference in the changes in REE, pulmonary function, and growth status over time has several possible explanations. It could be due to the known gender differences in the time of onset of puberty which may result in different patterns in this pre- to peripubertal age range. In addition, the girls in the control group declined somewhat in weight Z score, but only half as much as the girls with CF. Differences in physical activity between boys and girls might explain differences in the pattern of growth due to negative energy balance. Alternatively, these findings may be indicative of gender differences in a disease course that begins before or very early in pubertal development. Althogh boys and girls were similar in growth status and FEV1 in the final year of the study, these longitudinal patterns may have important implications for clinical care. In the clinical care of girls with CF with mild to moderate lung disease, pulmonary function measures may not be indicative of nutritional requirements, and the increasing REE may be likely to result in negative energy balance and growth failure if untreated.

This study was initiated before the identification of the locus for the CF gene, and therefore was not specifically designed to examine genotype differences in the longitudinal changes in REE and pulmonary function. Previous reports of the effect of genotype on REE have been inconsistent(11, 13, 14). Associations between theΔF508 homozygous genotype and pancreatic insufficiency are more consistently reported(13, 14, 42–45). With respect to pulmonary function, Kerem et al.(43) found better pulmonary function in patients with pancreatic sufficiency who were not homozygous for the ΔF508 allele. Ganet al.(45) found a greater rate of decline in pulmonary function in the ΔF508/ΔF508 adult group and in someΔF508 heterozygotes with severe mutations. At baseline in this study(11), there was no relationship between pulmonary function and genotype after adjusting for age, gender, and weight. In the present report, comparisons between ΔF508 homozygotes and other CF genotypes did not reveal consistent patterns in the longitudinal changes in growth, nutritional status, or REE, expressed as%pred WHO. In longitudinal models of FEV1 which included a nutritional status variable, theΔF508 homozygous state was associated with a decrease of 15 units of percent predicted FEV1 for children with the same%IBW. This new longitudinal finding requires further evaluations in other pediatric CF populations.

REE has been suggested as a more sensitive measure of clinical status than standard pulmonary function tests during short-term pulmonary exacerbations(46). The increase in REE in the girls with CF over the 3 y of this study with the concomitant decline in height Z score(and no measurable change in pulmonary function) strongly suggests that REE may be an important indicator of disease severity over longer time periods in young girls. It also suggests that the etiology of malnutrition and growth failure in CF may be different for boys and girls, especially as they approach and progress through puberty. A better understanding of these gender differences through further prospective investigations may improve the delivery of care to patients with CF and increase longevity.

In summary, this longitudinal study showed clearly that increased REE occurs in prepubertal children with CF with mild pulmonary disease. Furthermore, this study is the first prospective study to demonstrate that differences between children with CF and controls in REE increased over time, after accounting for changes in growth and body composition, and were not associated with declines in pulmonary function. The findings reported here are based on long-term changes in children with mild pulmonary manifestations of CF, further supporting REE as a sensitive measure of clinical status before development of moderate to severe lung disease. The different patterns of change with respect to pulmonary function and REE in male and female subjects need to be explored further and should be a consideration in future investigations of the relationships between REE, nutritional status, and pulmonary function. A simple measure of current nutritional status,%IBW, reflecting body weight for actual height regardless of previous stunting, was shown to be a good predictor of FEV1 over time. Further investigation of the relationship between%IBW and pulmonary function is needed to determine its prognostic value in a clinical setting. Genotype was found to be significant only in an explanatory model of prospective changes in FEV1, which included nutritional status. These findings demonstrate the need for sustained positive energy balance which results in optimal growth and adequate fat stores in children with CF. Clinically, we, like the CF Consensus Committee(47), suggest that the evaluation of growth, body composition and nutritional status are essential to the care of children and adults with CF. When increases in REE and decreases in growth and nutritional status are found, nutrition intervention is required even if these indicators are not associated with a measurable decline in pulmonary function.