CF carrier screening—making it meaningful
What the manufacturers offer for CF screening
Just over a year ago, obstetricians began offering to healthy women who were pregnant
or planning a pregnancy a genetic analysis of the CFTR gene, the gene
mutated in cystic fibrosis. Technically speaking, genetic analysis
for CF carrier status is a laboratory test, with the same considerations
as all laboratory tests. But a number of issues make this clinical
laboratory test unlike any other. “This is the first genetic
screening test of its kind,” says Elaine Lyon, PhD, medical
director of molecular genetics at ARUP Laboratories.
William Check, PhD
For example, a screening panel for CF carriers can include a wide
range of mutations, from one mutation with a detection rate of about
70 percent, to about 90 mutations with a detection rate of about 90
percent. Here the issue is not the typical tradeoff between sensitivity
and specificity, but a balance between sensitivity and technical feasibility.
What criteria guide such a decision?
Those detection rates, moreover, apply only to Caucasians of northern
European descent. The carrier detection rate for a 25-mutation panel,
for instance, ranges from about 55 percent for Hispanic Americans
to over 95 percent for Ashkenazi Jews; the rate for Asian Americans
is unknown. In how many laboratory tests does ethnicity play that
great a role?
Informing women about the test and counseling those who test positive
are other issues. An insufficient number of genetic counselors is
available for population-wide CF carrier screening. Before advocating
such a program, officials first had to determine whether obstetricians
could adequately inform and advise patients.
Then, too, for screening purposes, cystic fibrosis is sometimes not
really cystic fibrosis. Clinically, a condition called congenital
bilateral absence of the vas deferens, or CBAVD, which causes infertility,
is considered a mild congenital genital form of CF. However, geneticists
do not consider it severe enough to justify screening. CBAVD is often
associated with the presence in the CFTR gene of a polymorphism called
5T, but unless 5T is present with a particular combination of other
CFTR mutations, it does not cause CF. Yet some laboratories’
first-line testing panels include 5T. Is this appropriate?
Wayne Grody, MD, PhD, professor of pathology and laboratory medicine,
pediatrics, and human genetics at the University of California, Los
Angeles, School of Medicine, groups concerns about CF carrier screening
during the planning stages into three categories:
These concerns led to the American College of Medical Genetics
and the American Society of Human Genetics recommendation against
carrier screening in the 1990s, recalls Jean Amos, PhD, technical
director of molecular genetics at Specialty Laboratories. Dr. Grody
adds, “Some people even felt that CF is not uniformly serious
enough to justify screening.”
- Technical complexity of testing.
- Fear that because of this complexity there would not be enough
genetic professionals for the anticipated caseload.
- Clinical variability of CF.
Before CF carrier screening could become a reality, therefore, fundamental
concerns had to be addressed. “This is among the very first
large genetic carrier screening programs in the U.S. or the world,”
says Michael Watson, PhD, executive director of the ACMG and adjunct
professor of pediatrics at Washington University.
“Right now doing the test—the physical generation of
results—is a slam dunk,” says Dr. Amos. “But CF
carrier screening is different even from other high-volume molecular
genetic tests because it requires a complex interpretive report.”
Dr. Amos uses just three codes for all possible results of the Factor
V Leiden genetic test. “But for CF,” she says, “one
result can generate several reports,” depending on patient
Dr. Lyon agrees: “The technology to run this test is becoming
easier than dealing with the pre- and postanalytical phases. You
might think doing 25 mutations is complicated technically, and it
is. But when we set up the test we had technical validation done
within a reasonable time, then spent much longer wording reports,
getting ordering information correct, and providing forms and educational
Whereas medical genetics experts managed previous genetic tests,
CF carrier screening often lies in the hands of obstetricians, who
may not be familiar with the implications of such tests. “It
is important that clinicians who aren’t necessarily geneticists
be reminded in the laboratory report that results may be relevant
to other family members, as well as to the patient,” says
Hans C. Andersson, MD, of Tulane University School of Medicine.
Another complexity of genetic testing is that a mutation does not
equal disease, points out Dr. Andersson, who is associate professor
of pediatrics in the human genetics program and associate director
of genetic laboratories in the Hayward Genetic Center. “It
is easy to do mutations, but not easy to do interpretations,”
Before the discovery and cloning of the gene for CF in
1989, Dr. Grody says, “there was no way to do population screening
for CF carriers. It is a classic recessive disease. Carriers are
completely asymptomatic; even their sweat chloride is normal.”
First discovered was the ΔF508 mutation, a deletion of codon
508, which accounts for 70 percent of disease alleles in Caucasians.
Over a period of several years, more than 1,000 mutations accounting
for the remainder of the disease were discovered. “It was
not obvious how you would test for that many analytes,” Dr.
Grody says. It looked as though the highest practical detection
rate would be 80 to 85 percent. “That is less than the sensitivity
of most clinical laboratory tests,” Dr. Grody notes. A decade
of debate over whether to institute a carrier screening program
To break the impass, the National Institutes of Health funded five
pilot carrier screening programs. Results reported at a consensus
conference in 1997, along with other evidence, led the NIH expert
panel to recommend implementing prenatal and preconceptional carrier
screening and to select prenatal clinics as the preferred setting.
“Ideally we would like to do preconception screening,”
Dr. Grody says. That would allow options such as sperm donation
or in vitro fertilization with pre-implantation diagnosis. “But,”
he notes, “studies for many genetic diseases showed much lower
uptake for people not yet pregnant.”
Strongly influencing the panel’s recommendation was the good
outcome of Dr. Grody’s pilot study, in which genetic counselors
and trained volunteers approached patients directly in a prenatal
clinic. This strategy achieved about 85 percent uptake in an ethnically
diverse population, compared with 50 percent or less in other settings.
Dr. Grody also strongly credits the panel’s decision to the
success of an independent study by David R. Witt, MD, medical geneticist
and director of the regional CF prenatal screening program at Kaiser
Permanente Northern California. In 1991–1992, Dr. Witt conducted
a pilot study among more than 5,000 women, offering testing as part
of routine prenatal classes. A videotape and written brochure were
provided and, Dr. Witt says, “clearly made a dramatic improvement
in knowledge.” Of all eligible women, 78 percent decided to
undergo testing. Dr. Witt also evaluated patient attitudes and the
logistics of testing. His conclusions: “We found that you
can track large numbers of patients and specimens without making
mistakes, you can educate people orally, by written material, or
by videotape, and people generally are not upset about being identified
While the NIH panel’s 1997 recommendation for national CF
screening got things moving, it left many details unsettled. How
many mutations should be tested for? Should only some ethnic groups
be tested? Or all of them? Who should counsel patients? The American
College of Medical Genetics and the American College of Obstetricians
and Gynecologists established a joint committee in 1998 to answer
these questions. Not until 2001 were both organizations satisfied
enough to issue policy statements recommending CF carrier screening.
(For ACMG’s statement, see www.faseb.org/genetics/acmg/pol-32.htm.)
Selecting the mutation panel was particularly vexatious. Dr. Grody
chaired the subcommittee on laboratory testing, whose charge was
to select the mutations. “There was a big debate,” he
says. Eventually the group decided to include any mutation representing
0.1 percent or more of disease alleles in the CF Foundation’s
registry, a pan-ethnic population of 20,000 classical CF patients,
which yielded 25 mutations plus a few variants that modify the expression
of one of the mutations.
Dr. Grody calls that number “somewhat arbitrary.” The
incidence of many mutations included is right at the 0.1 percent
cutoff, he notes; some are so rare that there are no positive controls.
“I wouldn’t have minded a higher cutoff,” he says.
“In my view we could have had 16 mutations and it would have
worked just as well. People tend to lose sight of the fact that
this is just a screening test and we know we are going to miss some
carriers. Anything is better than what had been happening before.”
Sue Richards, MD, associate professor of molecular and human genetics
and director of the diagnostic sequencing laboratory at Baylor College
of Medicine, was also on the laboratory subcommittee. She recalls
that their work “dragged on” for more than two years.
“Garry Cutting [Garry Cutting, MD, of Johns Hopkins Medical
Institutions, a leading CF researcher] was instrumental in helping
develop perspective in selection of the mutation panel,” she
“There were both extremes,” Dr. Richards adds, “people
arguing for 40 to 50 mutations, which they were already doing, and
people who wanted a reduced panel. It makes sense to go for the
optimal effect—to do what you can at a reasonable cost. I
think we came up with a good number of mutations that are achievable
by many laboratories. At Baylor we had been doing screening for
more than 30 mutations for several years and we knew that you could
set that up on many platforms with existing technology.”
At the least, defining a specific set of mutations standardized
testing. In 1997, Dr. Amos says, “Laboratories were all over
the place, offering anywhere from one to 70 mutations.” And
having a standard panel enabled manufacturers to devise reagent
sets. (See “What the manufacturers
offer for CF screening,” pages 38–39.)
Whom to test was another major question. Significant variations
between ethnic groups are highlighted by updated data showing CF
incidence of one in 2,500 for non-Hispanic Caucasians, one in 15,000
for Hispanics, and one in 31,000 for Chinese Americans. With the
current mutation panel, carrier detection rates range from 97 percent
for Ashkenazi Jews to 90 percent for non-Hispanic Caucasians to
56 percent in Hispanic Americans. Nonetheless, screening is recommended
for all ethnic groups.
In the clinical service program Dr. Witt has set up at Kaiser Permanente,
screening is offered only to couples of which at least one member
is Caucasian. “There is debate among geneticists,” he
acknowledges, “about whether CF screening should be offered
to everyone.” He bases his decision on the decreased incidence
and decreased ability to detect carriers in Hispanics, African Americans,
and Asians. “I disagree very much with the way in which the
policy statements from ACMG and ACOG are phrased to ‘offer’
screening to Caucasians and to ‘make it available’ to
other groups,” he says. “I think that approach begs
the issue and doesn’t offer clear guidance. No wonder many
obstetricians are confused.”
Dr. Watson says ethnically limited screening is a “rational
approach to consider with the current panel. We are re-evaluating
the whole pa nel, in part because of ethnic issues.” It is
not surprising that the initial mutation panel is more sensitive
in Caucasians, since the incidence of CF is highest among Caucasians—one
in 25 are carriers. They have been the group studied most extensively.
Hispanics and African Americans carry mutations not found in non-Hispanic
Caucasians. “We now know that to effectively scan a multi-ethnic
population like in the U.S., we need more than 25 mutations,”
says Bernice Allitto, PhD, site director of Genzyme Genetics’
molecular diagnostic laboratory in Westborough, Mass. Dr. Allitto
and her colleagues showed that a panel of 64 mutations can detect
81 percent of African American and 72 percent of Hispanic carriers
(Genet Med. 2001;May–Jun; 3:168–176). Detection
among Asian Americans remains so low that Dr. Allitto does not give
a post-test frequency for that group. “There is not enough
information,” she says.
Of genetic counseling, Dr. Grody says, “There are clearly
not enough genetic counselors to handle the caseload of carrier
screening.” One goal of the pilot studies was to determine
whether other health care personnel could convey the necessary information.
“The answer was yes,” Dr. Grody says. One ACMG/ ACOG
subcommittee developed educational materials to train obstetricians
to do genetic counseling for this one disease. “ACOG, along
with ACMG, created a well-written booklet,” Dr. Grody says.
“Now obstetricians are supposed to be well enough informed
that they can handle questions. That part remains to be seen.”
One question remains: whether obstetricians have time in a routine
prenatal visit to convey the necessary details.
In an overall program, there is still a need for genetic counselors.
“Genzyme Genetics employs over 70 genetic counselors who see
patients in a clinical setting, in addition to the laboratory counselors.
One of the most important things genetic counselors do for us,”
says Dr. Allitto, who has six counselors on staff dedicated to the
molecular laboratory, “is to act as a liaison between the
laboratory and physicians.” Counselors provide information
to physicians who are non-geneticists, review patient samples for
adequate information to do appropriate testing, request and review
samples from family members if needed, and talk to physicians about
interpretation of testing outcomes. Genetic counselors call physicians
with positive results and explain the next step in the testing process.
About one-third of samples require followup for missing information
“that is critical for optimal result interpretation. That’s
a lot,” Dr. Allitto says, “and not atypical for reference
The situation at ARUP Laboratories is similar. “About two-thirds
of clients are sending us only partial or no clinical information,”
Dr. Lyon says. “That is very difficult.” A genetic counselor
calls a physician if the request is marked positive for family history
but does not indicate what that history is. Otherwise Dr. Lyon runs
the sample and provides a comprehensive report.
Interpreting and reporting the test results are proving
to be the most challenging parts of screening. Accurate interpretation
requires information on ethnicity and family history. Dr. Amos notes
that a negative result in a Caucasian of European descent with no
family history reduces the risk of being a carrier from one in 25
to one in 241. If that same person has a sibling with CF (and an
unknown mutation), a negative result reduces the risk for being
a carrier from two in three to one in six. For a Hispanic American
with no family history, a negative result reduces the risk from
one in 46 to one in 110. Sample laboratory reports for various interpretations,
along with a table of pre- and post-test risks by ethnicity, are
given in the appendix to the ACMG policy statement.
A more serious concern in reporting comes from the complex relation
between mutation pattern and CF phenotype. In the mutation panel,
along with 25 primary mutations, reflex testing was recommended
for a variant called 5T/7T/9T, a variable-length tract of thymidines
in the CFTR gene that modifies the expression of the R117H mutation.
When 5T is paired with R117H in a cis configuration and another
CF mutation is present in trans, the phenotype is variable. If the
other mutation is ΔF508, the person may have classic pulmonary
CF. When the other mutation is a “mild” mutation, the
overall clinical presentation is usually less severe. When 5T is
paired in a trans configuration with R117H and no other CFTR mutations
are present, it can be associated with CBAVD, as can 5T homozygosity.
However, 5T alone is rather common; it’s present in about
five to 10 percent of the population. “We don’t want
to pick up 5T homozygotes incidentally,” Dr. Grody says, “or
even 5T carriers in the absence of any other CF mutation. People
may want to prevent infertility in a male child, but that is not
why we have a CF screening program.”
To make matters more complicated, Dr. Watson says, “Despite
the fact that we recommend 5T as a reflex test only, the technology
went its own way to high-throughput, multiplex assays, and some
vendors and laboratories moved toward including 5T in the first-line
assays.” These laboratories felt obligated to give 5T status
with the primary result, even if no other mutation was present.
“I know of at least 16 cases in which women have undergone
amniocentesis, which has its own risk, for having this nondisease-causing
variant,” Dr. Watson says. “This suggests that their
physicians were not fully aware of the meaning of that result.”
If laboratories test for 5T in a first-line panel, they may have
no obligation to report it, in Dr. Watson’s view. “Many
laboratories indicated that lawyers told them if they tested for
it they had to tell patients the result. I am now soliciting a legal
opinion that will be included with the guideline revision.”
Dr. Watson draws an analogy to chemistry testing: If a physician
orders a sodium, the laboratory reports only that value, even though
efficiency dictates that they run the whole chemistry panel.
A second complication arose during the screening program with the
discovery that the mutation I148T was appearing about 100 times
as often as expected, based on its incidence among CF patients.
This is analogous to what had been shown in the screening pilot
studies with R117H and 5T. “What that told us,” Dr.
Watson says, “is that for I148T to cause disease, there must
be another mutation traveling with it.” Elizabeth Rohlfs,
PhD, and her colleagues at Genzyme identified a deletion (3199del6)
that is always present in CF patients with I148T. Because many testing
panels do not include 3199del6, laboratories should communicate
a positive result for I148T with caution. In the ongoing program
review, I148T can be removed and the deletion added, or both can
be included with one being reflex.
These and other recently recognized issues are discussed in a Genetics
in Medicine editorial by Drs. Watson, Grody, Richards, and others
To address the quality of current reporting, Dr. Andersson and Marie
Krousel-Wood, MD, MSPH, assistant dean in the Tulane University
Schools of Public Health and Medicine, requested anonymized reports
for the major CF mutation, ΔF508, and for Factor V Leiden from
all North American laboratories doing gene testing. “Most
genetic tests are developed in academic research laboratories,”
Dr. Andersson notes, “but they are offered by nongeneticist
clinicians, who are not always capable of fully interpreting and
utilizing the results.” A good report can help overcome this
problem. For both tests, they found that some laboratories provided
complete reports and others skimpy reports, with many in between.
“What stuck out for me,” Dr. Andersson says, “was
whether laboratories included a genetic counseling recommendation
for family members of patients who had a genetic abnormality. Nongenetic
tests refer only to the patient, but DNA-based tests can imply something
about the patient’s family members as well.” A significant
number of laboratories sent nothing about genetic counseling or
a reason to inform or refer in patients with an abnormality.
Drs. Andersson and Krousel-Wood next made three mock CF reports—one
complete, one intermediate, and one skimpy—and sent one report
to each of several hundred randomly selected physicians (pediatricians,
pulmonologists, and general practitioners), asking them to rate
the reports for completeness and perceived usefulness. “We
found that physicians rated more comprehensive reports as more useful
and were more satisfied with them, regardless of specialty,”
says Dr. Krousel-Wood. Completeness seemed to be related to action-oriented
items, such as a recommendation for genetic counseling. “I
would suggest that laboratories consider whether their form has
all the critical elements listed in our paper,” says Dr. Andersson
(Genet Med. 2002;4: 324–327).
Because of its extensive interpretive elements and its
requirement for genetic expertise, genetic testing for CF carrier
status should probably not be done in every laboratory. “This
is a test that is not appropriate for hospital laboratories,”
Dr. Amos says.
“While most molecular diagnostics laboratories have the capability
to perform and validate these tests,” says Dr. Grody, “I
would prefer to see it done in a laboratory that has some experience
testing for inherited disease, rather than a laboratory that does
only molecular microbiology.”
Dr. Allitto agrees that CF carrier testing is appropriate in a laboratory
that already does molecular diagnostics and genetic testing. “The
technology used for testing is straightforward, but the interpretation
of results is complex. CF is a disease with variable phenotype,
and we know that with mutations like R117H and I148T, followup testing
is necessary for accurate interpretation.” There is still
much to learn about genotype phenotype correlations for CF, she
says, “and getting the additional information to provide an
accurate interpretation is not typically a focus of a non-genetics
Ronald McGlennen, MD, medical director of the molecular diagnostics
laboratory and associate professor of laboratory medicine and pathology
at the University of Minnesota, would like to help more laboratories
get into genetic testing, including CF carrier screening. “Why
would a competent high-complexity laboratory or a reference facility
not get into gene-based testing?” he asks. One answer is that
much of the methodology for genetic testing, he says, “hearkens
back to the days of wet-bench chemistry. So some laboratories are
sitting on the fence wondering how we do gene-based testing.”
The motive for laboratories to take genetic tests in-house, in Dr.
McGlennen’s view, is that these tests command a high value
in the marketplace. “Laboratories that are going to take on
genetic testing will do it because of something like CF carrier
screening,” he says, “where there is a kind of mandate,
an instant volume that didn’t exist 12 months ago.”
To help such laboratories, Dr. McGlennen has formed a company, Access
Genetics, to provide an integrated kit combining nucleic acid extraction
and gene chemistry along with the necessary equipment and interpretation,
as well as consultation and support. He hopes to introduce this
service for CF screening in the first quarter of this year.
Demand for CF genetic testing has skyrocketed since the 2001 ACOG
statement. ARUP Laboratories, for example, saw an increase from
60 samples in January 2002 to 200 in February to 600 per month a
few months later. Nationally, Dr. Watson says of the recommendations,
“We made a huge impact, with 35 percent to hundredfold increases
in various laboratories. Some laboratories jumped from 1,000 samples
per month to 11,000 per month in 15 months.”
Even so, he says, “perhaps only 25 percent of the pregnant
population is being tested. We suspect that many clinicians are
not routinely making this a part of their daily practice.”
ACMG and ACOG are now surveying an ACOG practice group to determine
the level at which obstetricians are offering the test. “We
may need to take another shot at education,” Dr. Watson says.
If Dr. Watson’s suspicions are accurate, parents who were
not offered prenatal testing and had a child born with CF may soon
bring malpractice suits. Once ACOG officially recommended offering
screening, it became the standard of care.
Inequalities in testing may also be occurring, Dr. McGlennen suggests.
As he talks to laboratories and clinics, he sees a gradient of testing
rates—higher in affluent communities and fading in less affluent
areas. “In one part of town there may be 20 to 30 tests sent
out per week, while in another part of town there are zero,”
he says. He wonders if this anecdotal observation prevails across
Baylor’s Dr. Richards is concerned about screening’s
“programmatic” issues, which include following up on
positive cases and keeping a record of outcomes. One obstacle is
that, for insurance reasons, samples from both members of a couple
may not be sent to the same laboratory. The subsequent difficulty
in linking their test results may be compounded if they each see
a different physician. “The whole purpose of the program is
to identify carrier couples and do prenatal diagnosis on their fetus,”
Dr. Richards says. “So we need to be able to link both parents’
test results and get downstream outcomes.” Such a program
would be set up at the physician, as well as the laboratory, level.
Dr. Richards agrees that Dr. Witt’s program at Kaiser Permanente
is a role model for such a system.
Dr. Witt set up a service program in 1999 based on the sequential
model used in his pilot study: Testing is offered to a pregnant
woman first, and to her male partner only if she tests positive.
“Our rationale is that we only want to identify high-risk
couples, those in whom both partners are carriers,” Dr. Witt
says. High-risk couples are offered fetal testing by amniocentesis.
Dr. Witt says uptake has remained “very high”—about
70 percent of pregnant women get tested. When a woman tests positive,
about 90 percent of male partners get tested, he says. And about
90 percent of high-risk couples elect amniocentesis.
Among about 43,500 pregnant women screened to date, about 1,500
carriers were identified, resulting in roughly 45 high-risk couples.
At this point, Dr. Witt says, the situation gets more complicated.
“We need to distinguish severe and mild mutations,”
he says. In seven cases, the fetus carried two severe mutations,
such as ΔF508. In all seven cases, the parents elected termination.
Of several pregnancies in which the fetus had one severe and one
mild mutation, two were terminated and the others were continued
In all other cases in which both parents were carriers, amniocentesis
showed that the fetus was either unaffected or a carrier and pregnancy
was continued. “In these instances, the parents were very
reassured by having this information,” Dr. Witt says.
When one parent has a mild mutation and the other has a severe mutation,
sometimes, with proper counseling, the couple chooses not to have
amniocentesis, Dr. Witt says. “Or if they do have amniocentesis,
they don’t terminate,” he says. “Most [geneticists]
would say that a fetus with one mild and one severe mutation is
not at risk for classic CF, although it may be at risk for pancreatic-sufficient
CF, which is usually moderate, or perhaps just chronic bronchitis,
which some would not even call CF. But there is a certain amount
Dr. Richards calls the huge increase in CF testing that
the recommendations spawned “a good thing.”
“Now we just have to make sure we do it right and address
the clinical implications of these various mutations,” she
Whether CF carrier screening is done right could have consequences
beyond this program.
“There are several other equally complex genetic diseases
waiting in the wings as potential targets for population screening,”
Dr. Grody says, “but CF as the first mass screening program
at the DNA level is kind of a test case. Whether it succeeds or
fails may well determine whether or how quickly we move on to those
other disease targets.”
William Check is a medical writer in Wilmette, Ill.