Serum Linoleic Acid Status as a Clinical Indicator of Essential Fatty Acid Status in Children With Cystic Fibrosis

Background: Children with cystic fibrosis (CF) and pancreatic insufficiency (PI) are at increased risk for essential fatty acid (EFA) deficiency.

Objectives: To investigate serum markers of EFA status in children with CF and PI and their association with growth, body composition, and lung function.

Patients and Methods: Serum phospholipid fatty acid, growth, and forced expiratory volume at 1 second (FEV₁, percentage predicted) status were assessed at baseline and 12 months in 77 children with CF and PI, 7 to 10 years old. Longitudinal mixed-effects models were used to compare associations of the triene:tetraene ratio (ratio of eicosatrienoic acid to arachidonic acid) and serum linoleic acid (as a molar percentage of total serum phospholipid fatty acids, or mol%) with the clinical outcomes. Controls for serum fatty acid were 23 healthy white age- and sex-matched children.

Results: Children with CF and PI had higher median triene:tetraene ratio and lower linoleic acid than healthy controls. Depending on the triene:tetraene ratio cutoff point used (0.04 or 0.02), either 17% or 52% of the children with CF had EFA deficiency, respectively. Only linoleic acid was significantly and positively associated with z scores for weight, height, body mass index, upper arm muscle area, and FEV₁ at baseline. Children with linoleic acid at 21 mol% or higher had significantly better growth and pulmonary status than those with lower concentrations.

Conclusions: Serum phospholipid linoleic acid at 21 mol% or higher was associated with better growth, body composition, and FEV₁. No clinical outcome associations were found with the triene:tetraene ratio. These findings suggest that linoleic acid concentration was a more clinically relevant biomarker of EFA status than the triene:tetraene ratio in children with CF and PI. Further research is warranted to validate this specific percentage of linoleic acid cutoff point as a new recommendation for clinical use.

Humans lack the enzymes to desaturate fatty acids (FAs) at the third and sixth carbon from the methyl end of the molecule, and are dependent on dietary sources to prevent essential fatty acid (EFA) deficiency. Linoleic acid (LA, 18:2w6), and α-linolenic acid (ALA, 18:3w3) are the omega-6 and omega-3 EFAs in humans. LA is an important membrane constituent; both of these EFAs have multiple cellular structural and functional roles, and are precursors to long-chain polyunsaturated FAs, which by themselves and through prostanoid products, are growth factors and inflammatory mediators that also influence gene expression ^(1,2).

EFA deficiency has been described in patients with cystic fibrosis (CF) for more than 40 years ⁽³⁾, and many CF clinical symptoms have been shown to be influenced by EFA deficiency ^(4-7). EFA status is often clinically defined by the ratio of eicosatrienoic acid (ETA, also known as mead acid, 20:3w9) to the important tetraenoic acid arachidonic acid (AA, 20:4w6) in serum or plasma ⁽⁸⁾. This ratio is known as the triene:tetraene ratio (T:T), or the EFA deficiency index. Biochemical evidence of EFA deficiency may precede clinical signs in children with CF ⁽⁹⁾. LA (the omega-6 EFA) is also a biochemical indicator of EFA status. Decreased serum LA concentration is a frequently described EFA abnormality in infants, children, and adults with CF ^(10,11). Our study had 2 aims: to examine serum LA and T:T ratio status in subjects with CF and pancreatic insufficiency (PI) as compared with white age-matched healthy control subjects, and to assess clinical associations of serum phospholipid LA concentration and T:T ratio with growth, body composition, and pulmonary function in subjects with CF and PI.

PATIENTS AND METHODS

Children with CF and PI (ages 7-10 years) from 13 CF centers participated in a 24-month study of nutritional status and progression of pulmonary disease. A subset of these subjects elected to participate in a behavioral and nutritional educational intervention previously shown to be effective in boosting energy intake ^(12,13), in accordance with the CF nutritional recommendations.

At the 12- and 24-month visits, serum FA and additional data were collected, and these visits are referred to as baseline and 12-month data for the purposes of this analysis. An additional anthropometric assessment occurred at 6 months. The behavioral intervention occurred in a subset of subjects with CF approximately 12 months before the baseline serum, pulmonary, and anthropometric measurements that constitute this study. The diagnoses of CF and PI were made by the home CF center based on clinical symptoms and upon duplicate quantitative pilocarpine iontophoresis sweat test, with chloride values of >60 mEq/L. PI was diagnosed by 72-hour quantitative fecal fat collection and analysis of dietary intake, with <93% fat absorption and/or a stool trypsin value of <80 μg/L. Children were excluded if they had a forced expiratory volume at 1 second (FEV₁) <40% of predicted, significant liver disease, type 1 diabetes mellitus, Burkholderia cepacia sputum colonization, or other medical conditions or medications known to affect growth. Age- and sex-matched healthy white children who participated in a bone health study at The Children's Hospital of Philadelphia (CHOP) served as control subjects. These healthy children were recruited from the primary care practice at CHOP and affiliated pediatric practices. Exclusion criteria included use of any medication known to affect growth (eg, thyroxin, growth hormone, current or previous oral steroid medication), height or weight less than the third percentile for age, percent of ideal body weight >130%, and significant developmental delay or impairments. None of the control subjects had preexisting diagnosed gastrointestinal disorders or significant symptoms to suggest gastrointestinal, hepatic, or pancreatic disease. Both protocols were approved by the Committee for the Protection of Human Subjects of the Institutional Review Board at CHOP and at the subjects' respective home institutions. Written informed consent and age-appropriate assent were obtained from the parent or legal guardian and each subject, respectively. Subjects with CF were seen at the CHOP Clinical Translational Research Center for 2 (at baseline and at 12 months) overnight admissions and at their respective home CF centers for a day visit at 6 months, while in their usual state of good health. Evaluations at each visit included clinical status, anthropometry, and dietary (food and supplement) intake. Additionally, subjects with CF had phlebotomy, spirometry, and fecal collections performed at the baseline and 12-month visits. The healthy control subjects had 1 baseline assessment performed including phlebotomy and anthropometry during a 1-day study visit. For the healthy control subjects, only the baseline data were included for this study.

Energy Intake

Seven-day home-based weighed food records were obtained from children with CF at baseline after verbal and written instructions were provided along with measuring cups and spoons and digital food scales. Research dietitians analyzed the diet records (Nutrition Data System, Minneapolis, MN). Details of the specific brands of supplements, frequency, and dose were recorded. Total grams of fat; kilocalories from fat, carbohydrates, and protein; and total energy consumption were calculated. The estimated energy requirement-adjusted for age, sex, height, and weight, as well as for physical activity in the active range as determined previously-also was calculated ⁽¹⁴⁾. LA and ALA intake were measured in grams per day and calculated as a percentage of the adequate intake level (AI) of the dietary reference intakes ⁽¹⁵⁾.

Serum Fatty Acid Analysis

For the subjects with CF, fasting serum phospholipid FA status was determined at the baseline and 12-month visits. A single, nonfasting serum phospholipid FA assessment for an age- and sex-matched sample of healthy control subjects at the baseline visit was used for comparison. Total lipids of serum were extracted according to Folch et al ⁽¹⁶⁾. Serum phospholipids were fractionated on a single Sep-Pak amino propyl cartridge (Waters, Milford, MA), subsequently separated by capillary gas-liquid chromatography as previously described ⁽¹⁷⁾, and recorded with Chem Station software (Hewlett Packard GC, Wilmington, DE). The FA methyl esters were identified and quantified by comparison with pure reference substances (Sigma Aldrich Sweden, Stockholm, Sweden), with heneicosanoic acid (21:0) as the internal standard. The fatty acids LA, ALA, AA, ETA, and oleic acid (OA) are reported here. All of the values are expressed as molar percentage of total serum phospholipid FA (mol%). There is no accepted standard for defining EFA status; therefore, EFA status was explored using different, previously published clinically used cutoff points in studies in subjects with CF for the T:T ratios of 0.02 or less and 0.04 or less ^(18-20), and from 20 to 26 mol% for serum LA ^(9-11,21-29).

Body Composition, Growth, and Clinical Status

Height and weight were measured using standard techniques ⁽³⁰⁾ with a stadiometer accurate to 0.1 cm (Holtain, Crymych, UK), and a digital scale accurate to 0.1 kg (Scaletronix, White Plains, NY). Height was adjusted for genetic potential ⁽³¹⁾ from measured or reported biological parent heights for the subjects with CF. z Scores for height (HAZ), adjusted height (adjHAZ), weight (WAZ), and body mass index (kg/m²) (BMIZ) also were computed ⁽³²⁾. Mid-upper arm circumference was measured using a flexible plastic measuring tape (Ross Laboratories, Columbus, OH). A skinfold caliper (Holtain) was used to measure triceps and subscapular skinfold thickness on the right side. Upper arm muscle and fat areas were derived ^(33,34), and z scores for upper arm muscle area (UAMAZ) and upper arm fat area (UAFAZ) were computed ⁽³⁴⁾. These determinations were made for the subjects with CF at the baseline, 6-, and 12-month visits. Height, weight, and body mass index (BMI) measurements were analyzed for the healthy control subjects for the baseline visit only.

Pulmonary function was evaluated by standard methods for spirometry ^(35,36) following inhaled albuterol and chest physiotherapy in the children with CF at the baseline and 12-month visits. FEV₁ percent predicted was used as the measure of pulmonary function ⁽³⁷⁾, and is presented using both the equations by Knudson et al ⁽³⁸⁾ and by Wang et al ⁽³⁹⁾.

Data Analysis

Descriptive analysis was performed using means, standard deviations, medians, minimum and maximum values for continuous variables, and frequency distributions for categorical variables. Descriptive statistics and exploratory graphing techniques were used to assess the normality of the continuous variables. Demographic, growth, pulmonary status, and select serum FA comparisons between children with CF and healthy controls were assessed using Student t test for independent samples, or Mann-Whitney-Wilcoxon tests for non-normally distributed variables, because sample sizes were unequal. Associations between dietary intake and serum FA data were measured for subjects in both groups for whom both complete dietary data and serum FA status were available. The associations between dietary LA intake, serum FA, and growth, nutritional, and pulmonary status were measured by Pearson correlation coefficient for normally distributed variables or Spearman rho coefficients for non-normally distributed variables.

T:T cutoff points were derived from those published in the literature ^(18-20). The selection of the cutoff points for LA was driven by our data and by those most frequently used in the literature (26 mol%) ^(10,11,27). The LA values for our CF subjects were between 15 and 29 mol%; 12% of the observations had values <19 mol%, 10% had values >26 mol%, and 78% of the observations (n = 60) were within the values of 19 to 26 mol%. Therefore, we explored 7 different LA cutoff points at baseline (from 20 to 26 mol% in increments of 1; ie, <20 vs ≥20, <21 vs ≥21, etc, to <26 vs ≥26 mol%) using mixed-effects models for longitudinal data ⁽⁴⁰⁾. Because this is a secondary data analysis of a dataset of modest size, our analyses were exploratory. For each LA cutoff point, 4 models were fitted. Regression coefficients, standard errors, P values for each predictor within the models, and a measure of relative goodness of fit (Akaike information criterion [AIC]) ⁽⁴¹⁾ were obtained. Our goal was to explore cutoff points with statistically significant P values across the 4 outcomes (adjHAZ, WAZ, UAMAZ, and FEV₁). These models used all available information for each subject; specifically, at baseline, 6, and 12 months for growth and body composition, and at baseline and 12 months for FEV₁. The predictors were the FA status at baseline, time, and the FA status by time interaction. In addition, sex was included in all models. The data were evaluated for the cutoff point with the higher statistical significance and lower AIC value across the 4 outcomes (adjHAZ, WAZ, UAMAZ, and FEV₁). The type I error rate was adjusted by using the Tukey-Ciminera-Heyse adjustment for multiple comparisons for moderately related outcomes ^(42,43); the P value was set as P = 0.010. Agreement between the best cutoff value (≥21) of serum percentage of LA, obtained from our mixed-effects models, and 2 cutoff values of the T:T ratio (≤0.02 and ≤0.04) was evaluated by using the κ coefficient. All of the tests were 2-sided and were performed with Stata 8.0 (StataCorp, College Station, TX) or SAS version 9.1.3 (SAS, Cary, NC).

RESULTS

The study included 77 subjects (51% female) with CF and PI of mean age 8.4 ± 0.9 (7.0-10.0) years and 23 healthy control subjects (57% female) of mean age 8.4 ± 1.1 (7.2-11.0) years. Comparisons between subjects with CF and controls were based on cross-sectional data and were restricted to the baseline visit for both groups. The subjects with CF did not differ from controls with respect to age, sex, and growth status (Table 1). Subjects with CF had lower serum LA and AA-and higher serum ALA, OA, and ETA mol%-compared with control subjects. T:T ratio was higher for children with CF as compared with the controls (0.029 ± 0.030 vs 0.011 ± 0.004; P < 0.001). All of the control subjects and 48% of children with CF had a T:T ratio <0.02, and 17% of the children with CF had a T:T ratio >0.04. Sex differences were seen for ETA and T:T status in the subjects with CF. The female subjects had lower median serum ETA mol% (0.151 [0.055-0.802] vs 0.198 [0.060-1.560]; P = 0.017) than the male subjects with CF. Males had higher median T:T than the female subjects with CF 0.024 (0.008-0.219) vs 0.018 (0.004-0.137), P = 0.015.

Table 1

Dietary data were available for 86% of subjects with CF and for 96% of the healthy control subjects. The subjects with CF had greater energy, fat, and LA intake than the healthy controls. Of subjects with CF, 26% had LA intake <100% AI, and 21% of subjects had ALA intake <100% AI (Table 1). For both groups, LA, ALA, and total energy unit intake were not associated with serum LA or ALA status.

Associations between serum phospholipid FA, T:T ratio, and growth, body composition, and FEV₁ are shown in Table 2 for the subjects with CF. The LA status was significantly and positively associated with growth status and UAMAZ. The association of serum LA status with FEV₁ percent predicted (both Knudsen and Wang methods) did not reach statistical significance in this model (P < 0.01); however, a trend was observed (P < 0.05). The T:T ratio was not correlated with any of the outcomes of interest. The relations between ALA, OA, AA, and ETA and clinical measures are shown in Table 2. ALA was positively and OA negatively associated with growth status. AA was negatively associated with FEV₁ (Wang method only). Similar correlations between these clinical measures and LA, ALA, AA, OA, ETA, and T:T ratio were observed for the 12-month data, with the exception of the association between percent of LA and FEV₁ (not shown).

Table 2

Table 3 presents the results from the longitudinal mixed effects analyses corresponding to the 3 selected percentage of LA cutoff points (≥20, ≥21, and ≥26), and 2 T:T ratio cutoff points (≤0.02 and ≤0.04). There were no statistically significant associations between T:T ratio groups and growth, body composition, or pulmonary function. Growth status and FEV₁ differed significantly between groups defined by percentage of LA. Even though some significant group differences were observed at every cutoff point, the percentage of LA associations for every clinical outcome of interest was obtained only for the cutoff point ≥21% of LA. Compared with the group of children with LA <21%, children with LA ≥21% had significantly better WAZ (0.87 z score; P = 0.002; AIC 301.4), better UAMAZ (+0.93 z score; P = 0.002; AIC 426.0), and higher FEV₁ (12.9% predicted; P = 0.006; AIC 1172.0). Results did not reach statistical significance for the adjHAZ (0.67 z score; P = 0.012; AIC 164.0). The results using an LA cutoff point of 20% are presented to illustrate the stepwise increased associations for growth, nutritional status, and FEV₁ as serum LA increased from 20% to 21%. When children were divided using 26% LA as the cutoff point (the most frequently used cutoff point in the literature ^{(10,21,25-29)}), those with 26% or higher LA had the best growth status; however, the small sample size of this group limited the power to detect statistical significance. Sex effects were not seen in any of the models. Similar longitudinal mixed effects models were performed for BMIZ and UAFAZ, and the results were not significant (not shown). Although percentage of LA status was associated with growth, body composition, and FEV₁ at any given point in time, it did not predict change in these clinical outcomes during 1 year because the group × time interaction was not significant in any of the models.

Table 3

Potential differences in FA status were explored between subjects with CF and PI participating and those not participating in the behavioral intervention to boost energy intake 12 months before the baseline in this study. Of the subjects, 25 (32%) participated in this intervention, and did not differ from those who did not participate with respect to age, sex, serum FA status (LA or T:T ratio), or pulmonary function at baseline. The associations of serum FA status with growth and pulmonary status were similar in both groups at baseline and 12 months, and in the linear mixed-effects models did not alter the significant effects of serum LA status predicting growth and pulmonary outcomes.

The relation between T:T ratio and serum phospholipid percentage of LA for the children with CF at baseline is shown in Figure 1. Using the Spearman rank correlation, percentage of LA and the T:T ratio were significantly and negatively associated (r = -0.623, P < 0.0001). Serum LA was 26% or more of total serum FA in only 10% of children with CF, and 42% had serum LA <21%. Only 1 of the 45 children with LA 21% or higher had a T:T ratio >0.04, whereas 38% of children with LA <21% had a T:T ratio >0.04 (χ² = 16.59; df = 1; P < 0.001; κ = 0.39). When comparing an LA cutoff point of 21% to T:T ratio of 0.02, 38% of children with LA 21% or higher had a T:T ratio >0.02, whereas 72% of the children with LA <21% had a T:T ratio >0.02 (χ² = 8.71; df = 1; P < 0.0013; κ = 0.39).

Fig. 1

DISCUSSION

The incidence of EFA deficiency was high in children with CF and PI; depending on the T:T ratio cutoff used (0.04 or 0.02), EFA deficiency was present in 17% or 52%, respectively, of these preadolescent subjects. Percentage of LA and AA were lower and ETA was higher in subjects with CF compared with the control subjects. These data are the first reported to demonstrate that percentage of LA status was correlated with FEV₁ in children ⁽⁴⁴⁾, in contrast to previous studies ^(45,46). LA status was correlated with important clinical CF outcomes of growth, muscle stores, and pulmonary outcomes, whereas the currently recommended T:T ratio method of EFA status assessment ⁽⁴⁷⁾ was not.

EFA deficiency has been described frequently in subjects with CF ^(3,10,48-50). The most commonly reported polyunsaturated FA abnormality reported in CF is decreased LA status compared with healthy control subjects ^(51,52). The classical clinical manifestations of EFA deficiency include desquamating skin rashes, alopecia, easy bruising, clotting disorders, and impaired growth ^(53-55). FA abnormalities and EFA deficiency have been reported in infants with newly diagnosed CF ^(11,56), malnourished children with CF ⁽⁵⁶⁾, and in well-nourished children ^(8,9) with CF. The T:T ratio (ETA:AA) also has been reported to be abnormal (ie, elevated) in studies in subjects with CF ^(51,52). Our aim was to identify a biochemical indicator of EFA deficiency that was related to important clinical indicators in children with CF, such as growth, body composition, and/or pulmonary status. Although these outcomes are influenced by many factors in CF, they also may be additional subtle, subclinical signs of EFA deficiency.

Studies conducted in subjects with CF have identified multiple factors that influence serum LA status, including dietary fat intake and absorption ^(3,27,57), energy intake and energy balance ^(24,25,58), and increased FA turnover. When energy intake is inadequate to meet energy needs (negative energy balance), LA may be used for energy, and this may result in decreased LA body stores and status ⁽⁵⁹⁾. Low LA status has been reported in subjects with CF with normal fat absorption, as well as in subjects without clinical signs of EFA deficiency ^(60,61). Increased FA turnover has been shown to be associated with greater AA membrane release ^(62-65), with LA deficiency resulting in increase of proinflammatory eicosanoids ^(66,67). In contrast, EFA deficiency in non-CF subjects has been associated with decreased eicosanoid production ^(68-71). These derangements in FA metabolism in CF may relate to the primary defect ^(8,72) and underscore the complexities of FA abnormalities in subjects with CF, which cannot be entirely explained by dietary intake and fat malabsorption alone. FA metabolism abnormalities observed in CF may be of additional biological significance for this specific population.

In healthy populations, the incidence of EFA deficiency is rare, and the dietary reference intakes ⁽¹⁵⁾ define LA adequacy based on typical LA intake in these populations. LA and EFA deficiency have been traditionally described by the presence of the physical findings in deficiency states and/or by biochemical evidence (ie, an abnormal T:T ratio). The physical findings of EFA deficiency in children and adults are from enteral and parenteral energy unit supplementation studies in which fat was either limited or withheld ^(53-55,73). The biochemical approach to defining EFA deficiency is by an abnormal T:T ratio, which represents a shift in the biosynthetic desaturation and elongation pathways for the omega-3, -6, and -9 FAs ⁽⁷⁴⁾. These changes occur when serum ETA is increased and/or when serum AA is decreased ⁽⁴⁹⁾, resulting from diminished LA and ALA as compared with other FA such as OA in healthy subjects. The data from the present study and from the literature illustrate this shift in metabolism resulting in elevated T:T ratio with decreased LA status (Fig. 1).

The T:T ratio varies by geographical location and by dietary intake patterns across healthy human populations. What constitutes typical, or within the normal range, FA status can change over time for a population and/or across populations, and is exemplified by the evolution of the definition of a normal T:T ratio. The original T:T ratio of >0.04 denoting deficiency was based on the work of Holman from 1960 ⁽²⁰⁾. Subsequent T:T ratio investigations in healthy populations with normal omega-6 FA status by Holman and others demonstrated a T:T ratio of 0.010 ± 0.008 (mean ± standard deviation [SD]), leading to the recommendation that T:T >0.02 be considered abnormal ^{(18,23,28,75)}. This threshold has been adopted by some CF investigators ⁽²⁷⁾. The T:T ratio from our healthy control subjects was consistent with this revised T:T ratio; none had a T:T ratio >0.02. Similarly, geographical and population-specific dietary variations in LA intake may influence LA status ^(76,77), thereby influencing what is considered normal percentage of LA for different healthy groups.

The challenge in defining normal FA status in subjects with CF is in determining whether CF recommendations should be limited to a comparison with a healthy population or should reflect a CF-specific clinically meaningful cutoff point. The associations of LA status with growth and nutritional status as previously reported and as presented in the present study suggest the cutoff point to define adequate EFA status in CF could be developed on the basis of health outcomes, as opposed to the methodology used to define EFA adequacy in healthy populations detailed above.

The manner in which biochemical polyunsaturated FA status is reported is relevant to this discussion. LA status is most often reported in mol%, and conveys information regarding the physical properties and functional aspects of cell membranes ⁽⁷⁸⁾. Reporting FA status by absolute amounts (such as μg/dL) has been used less frequently in the CF research and clinical care literature. In addition, absolute FA concentration varies by age and does not consistently relate to mol% ⁽⁷⁹⁾. In the CF literature, reference values for serum phospholipid LA vary from 20% or higher to 27% or higher mol% total FA, with LA 26% or higher as the most frequently used value ^{(17,21,22,24-28)}. In the present study, LA 21% or higher was an informative cutoff point for growth and pulmonary status associations in children, with a threshold effect observed. In addition T:T and percentage of LA are biochemically related, and LA 21% or higher conveyed similar biochemical EFA status information as T:T 0.04 or less.

This study explored the relation between serum EFA status and clinical measures as a secondary analysis of 2 studies, and this was a limitation. Because longitudinal data were available for the subjects with CF and not for the healthy controls, the predictive value of EFA status at baseline on future growth status was explored only in the subjects with CF. The possible differences in phospholipid FA profiles in fed/fasted states and with respect to diurnal variation are, in part, mitigated by the absence of reports suggesting such variations occur. In addition, an extensive body of literature suggests children with CF have lower serum LA status and EFA deficiency than healthy children. The methodology-based potential differences in serum FA status between these 2 groups does not alter the key findings of the relation of serum FA status to clinically relevant outcomes in the children with CF. Last, because model building and hypothesis testing need to be conducted on separate data sets for hypothesis testing to be unbiased, further research is warranted to validate this specific percentage of LA cutoff point as a new recommendation for clinical use at this time.

In summary, these findings suggest that serum LA status was a better indicator of EFA status in children with CF and PI as compared with the T:T ratio. Depending on the T:T ratio used (0.04 or 0.02), 17% or 52%, respectively, of the subjects with CF had EFA deficiency. Although an abnormal T:T ratio is indicative of a shift in FA metabolism due to EFA insufficiency or deficiency, it was not related to the selected outcomes including growth and pulmonary status. In the present study, LA status was associated with these CF-specific clinically important outcomes. LA 21% or higher was an informative marker. Prospective studies are required to confirm and expand these findings before making recommendations for use of this specific percentage of LA cutoff point as a change in standards of clinical care. As an interim step, we suggest that LA status be determined in addition to the T:T ratio for assessment of EFA status in children with CF and PI, in the development of optimal nutritional care of patients with CF.

Acknowledgments

We would like to thank the participating children and families, the cystic fibrosis centers that have made this study possible, and the Nutrition and Growth Laboratory staff at the Children's Hospital of Philadelphia. Participating CF centers included Albany Medical Center, Albany, NY; Children's Hospital of Buffalo, Buffalo, NY; Children's Hospital of Philadelphia, Philadelphia, PA; Children's Medical Center Dayton, Dayton, OH; Children's National Medical Center, Washington, DC; Egleston Center at Emory University, Atlanta, GA; Hershey Medical Center, Hershey, PA; Johns Hopkins Children's Center, Baltimore, MD; Long Island College Hospital, Brooklyn, NY; Schneider's Children's Hospital of Long Island, New Hyde Park, NY; St Christopher's Hospital for Children, Philadelphia; State University of New York, Stony Brook; and University of Florida, Gainesville.