Use of Formula Fortified With DHA in Infants With Cystic Fibrosis

This study brings together several areas of CF and non-CF research: the finding that CF
knock-out mice exhibit an abnormality in AA/DHA ratio in membrane bound fatty acids from
tissue which expresses CFTR; the fact that very high doses of DHA fed to CF knock-out mice
can correct many of the abnormalities seen in these animals; research on fatty acid content
in human breast milk; the effects of breast feeding in CF; and the fact that newborn
screening for CF is becoming more widespread and may allow for therapeutic interventions very
early in life. This is the first study of which we are aware to look at a therapeutic
intervention in children diagnosed with CF by newborn screening.

Fatty Acid Metabolism in CF It has been recognized for years that patients with CF have
abnormalities in their fatty acid profile.(1) Initially, this was felt to be secondary to
malabsorption of essential fats. However, in 1986, Strandvik's laboratory proposed that an
abnormality in fatty acid turnover (specifically arachidonic acid metabolism) was a primary
problem in patients with CF(2). More recently, Freedman et al (3) have shown that CFTR
knockout mice have an abnormality in membrane bound long chain polyunsaturated fatty acids in
CFTR expressing tissue with an increased ratio of arachidonic acid (AA) to docosahexanoic
acid (DHA) compared to control animals. They have also shown that therapy with formula
fortified with high doses of DHA reverses the lipid abnormality and ameliorates the
pancreatic duct changes seen in these mice and decreases the inflammatory response to inhaled
lipopolysaccharide(3,4).

Freedman et al (5) have gone on to show that abnormalities in membrane bound fatty acids in
CFTR expressing tissues in humans are similar to that seen in CF knockout mice. They and
Strandvik et al(6) have shown that this fatty acid abnormality is dependent on genotype with
more severe fatty acid abnormalities found in patients with "severe" mutations (mutations
associated with pancreatic insufficiency). That this abnormality is a primary part of the
disease and not secondary to malabsorption is supported by the fact that obligate
heterozygotes have fatty acid abnormalities intermediate between affected individuals and
normal controls(5). In summary, fatty acid abnormalities appear to be a primary defect in CF
and are directly related to the patient's genotype. Furthermore, dietary correction of this
fatty acid imbalance improves symptoms in mice. Previous human studies looking at correction
of fatty acid imbalances in CF patients have focused on the 18 carbon precursors to AA and
DHA (reference 7, for example) or have used very brief treatment periods with
eicosapentaenoic acid (EPA) (which was not effective in Freedman's mice) in individuals who
already have established disease (8).

Human Breast Milk and Breast Feeding in CF Human breast milk contains low levels of DHA but
standard infant formula does not (9).

Traditional infant formulas such as Similac and Enfamil have 18 carbon fatty acids as their
source of long chain polyunsaturated fatty acids. These include linoleic acid (18 carbons, 2
double bonds, the last double bond 6 carbons from the methyl end; 18:2n-6) and linolenic acid
(18:3n-3). Eighteen carbon fatty acids may be desaturated and elongated to make AA (20:4n-6)
and DHA (22:6n-3) which are then incorporated into membrane phospholipids (10). Alternately,
these 18 carbon fatty acids can be 0-oxidized in mitochondria and used as an energy source.
AA and DHA are necessary for brain growth.(11) The most rapid period of brain growth is the
third trimester and it seems likely that this is the most important period of time for
accumulation of these fatty acids by the fetus. Prematurely born infants miss out on
placental transfer of AA and DHA in the last trimester. They are also more likely to need to
use long chain polyunsaturated fatty acids as an energy source since they may have limited
caloric intake and increased caloric expenditure due to respiratory disease. It is not
surprising, therefore, that AA and DHA supplemented formulas would be found to have a more
profound effect upon pre-term than term infants. In fact, new formulas containing AA and DHA
have been found to improve neurodevelopment in premature infants (12). However, there is
controversy as to whether or not there are neurodevelopmental advantages to term infants
being fed fatty acid supplemented formula with some authors reporting a benefit (13) and
others (including a very large, double blind, randomized study(14) not finding any difference
between term infants fed standard formula and those given long chain polyunsaturated fatty
acid supplemented formula. Since brain tissue does not express CFTR and brain levels of AA
and DHA were not abnormal in Freedman's mouse experiments (3), it seems unlikely that full
term humans with CF would have neurodevelopmental problems related to decreased brain DHA
levels different from the general population.

There is a limited reserve of n-3 fatty acids in tissues resulting in faster onset of DHA
than AA depletion in infants who have a limited intake of fatty acids. This could be
exaggerated in infants with CF in CFTR expressing tissue which is already low in DHA. Dietary
AA and DHA are preferentially acylated into tissue structural lipids, whereas dietary 18
carbon n-3 and n-6 fatty acids can be more readily used as an energy source (10). Thus for
some infants (premature infants and possibly CF infants) the longer chain fatty acids may be
essential dietary components.

The 20 and 22 carbon n-3 fatty acids (EPA and DHA) inhibit 0-6 desaturase and reduce
synthesis of AA from linoleic acid (18:2 n-6). (10) Therefore, unbalanced addition of DHA to
infant diets may result in abnormal tissue fatty acid composition. Taking advantage of this
desaturase inhibition by feeding infants a formula with DHA but without AA might be
beneficial for infants with CF assuming that they have intrinsically elevated AA tissue
levels and decreased DHA levels. However, such a strategy has not been tested and since there
is a normal AA/DHA ratio in non-CFTR expressing tissues this strategy might put CF infants at
risk of abnormal brain development by altering normal neural AA/DHA ratios. It seems prudent
to provide both AA and DHA to CF infants in levels consistent with those found in human
breast milk.

There is a great deal of variation in human breast milk long chain polyunsaturated fatty acid
content depending on the mother's diet with DHA levels ranging from as little as 0.2% of
fatty acids in women on vegetarian diets to 2.78% of fatty acids in Chinese women on a marine
diet (9,15). DHA has been granted GRAS status ("generally recognized as safe") by the FDA,
clearing its way for addition to infant formulas. Recently, Mead Johnson and Ross
Laboratories have released infant formulas enriched with long chain polyunsaturated fatty
acids, specifically AA and DHA. Many studies have demonstrated the safety of these formulas
(13,14,16,17).

Given that breast milk contains low levels of DHA and traditional formula does not, it is
reasonable to think that breast feeding might be beneficial for children with CF. Very few
studies have looked at this question. In the 1960s and 1970s several authors warned against
breast feeding because it was felt to precipitate hypoproteinemic, edematous states (18,19)
By 1990 77% of CF Centers were recommending breast feeding for infants with CF (which means
23% were not). (20) In 1991 Holliday et al (21) were able to demonstrate improved growth in
CF infants who were breast fed compared to those who were not. This study showed that breast
feeding was safe for infants with CF and possibly beneficial.

A recent nationwide survey has demonstrated an association between breast feeding and
improved outcome in CF individuals (22). Whether this is due to DHA, other constituents of
breast milk not found in commercial formula, or social factors is not clear. Related to this
survey result is the observation that some CFTR knockout mice do not develop pancreatic
insufficiency and CF gastrointestinal symptoms until they are weaned from their mother's milk
(Steven Freedman, personal communication). Mouse breast milk, like human milk, contains DHA.
Although 15% of humans with CF are born with obvious pancreatic insufficiency and bowel
obstruction (meconium ileus), many are not symptomatic on day one of life. Pancreatic
insufficiency appears to be a continuous, not discrete, variable with change in pancreatic
sufficiency status over time seen in many patients. In one study, only sixty percent of
infants with CF diagnosed using newborn screening were noted to be pancreatic insufficient at
diagnosis as determined by 72 hour fecal fat studies (23). By 12 months of age fat
malabsorption was seen in 92% of this cohort. These results are similar to those of Waters et
al (24) who also saw increasing prevalence of pancreatic dysfunction over time. It is
speculated that early intervention with a DHA containing formula might slow the inflammatory
destruction of the pancreas seen in humans over the first year of life.

LIPIL, manufactured by Mead Johnson has 0.32% fat as DHA. Mead Johnson has made a formula
with three times this amount of DHA for this study. The study formula, designated LIPIL x 3,
has 0.96% of fatty acids as DHA and retains the same AA concentration (0.64%) as the
commercial formula. This formula meets the FDA GRAS designation and does not require an IND
for use in a clinical trial.

Human infants have thrived for decades with the standard formulas that have been available;
it remains unclear if term infants derive benefit from the long chain polyunsaturated fatty
acids supplied by the newer formulas. Evidence indicates that term infants do well with
either conventional or fatty acid supplemented formula so there is no ethical dilemma to a
double-blind comparison of these formulas.

Newborn Screening for CF Many states have begun performing newborn screening for cystic
fibrosis. For example, Massachusetts began screening for CF in February 1999. This program
utilizes both IRT measurement and 39 mutation analysis. This has proved to be very good for
detecting children with CF early in life, the median age at diagnosis being 16 days (Anne
Commeau, personal communication). Many other states have begun performing newborn screening
for CF using a similar algorithm or an IRT/IRT algorithm in which two blood samples are
analyzed for IRT days apart. Compared to children diagnosed by conventional methods, Farrell
et al (25,26) have shown improvement in the nutritional status of CF patients detected by
newborn screening. It is believed that improved nutrition will lead to improved pulmonary
function later in life, but to date this has not been proven. In fact, the finding that it is
possible that newborn screening could lead to earlier acquisition of Pseudomonas aeruginosa
has worried some clinicians about the utility of newborn screening for this disease (27,28).
Thus, CF can be detected early in life using newborn screening and this has been shown to
lead to nutritional and possibly pulmonary benefits over time.

It is the goal of this study to see if a simple intervention early in life in those children
diagnosed with CF via newborn screening can make a difference in the progression of their
disease. It is hypothesized that if humans respond to DHA in a similar fashion to CFTR
knockout mice, then supplementation of these infants' diets with DHA early in life will
provide protection from pancreatic disease and pulmonary inflammation.

Inclusion Criteria:

- Infant diagnosed with CF and enrolled by 56 days of life

- Parental consent obtained

Exclusion Criteria:

- History of meconium ileus at birth that is resolved without surgical intervention (ie
enema)

- History of bowel resection for any reason

- Breast feeding

- Premature birth (<34 weeks gestation)

- Severe cholestasis (Direct Bilirubin > 2x upper limit of normal for age)

- Severe hypoalbuminemia (Albumin < 2.5 gm/dl)