College of American Pathologists
CAP Committees & Leadership CAP Calendar of Events Estore CAP Media Center CAP Foundation
 
About CAP    Career Center    Contact Us      
Search: Search
  [Advanced Search]  
 
CAP Home CAP Advocacy CAP Reference Resources and Publications CAP Education Programs CAP Accreditation and Laboratory Improvement CAP Members
CAP Home > CAP Reference Resources and Publications > CAP TODAY > CAP Today Archive 2001 > Sooner or later? Down screening debate
Printable Version

  Feature Story

title
 

cap today

Sooner or later? Down screening debate

December 2001
Anne Paxton

Understanding the disorder and the markers

Down syndrome, or trisomy 21, is a fetal disorder that is full of apparent contradictions. It is genetic, yet for the most part not inherited. It is the least lethal of the chromosome disorders, yet the most intensely studied. While the number of babies born with Down syndrome has fallen by half in the last two decades, the number of fetuses with Down has risen. And while the syndrome can be diagnosed with 100 percent certainty through well-established clinical lab testing, most prenatal screening research has focused on screening markers that will detect as few as six out of 10 Down syndrome cases.

The problem is that while two diagnostic tests with a 100 percent detection rate exist—amniocentesis in the second trimester of pregnancy and chorionic-villus sampling in the first—they are not only expensive but also invasive, and they carry a significant risk of causing miscarriage.

That has increased the stakes of screening markers that can at least detect the risk of Down syndrome earlier, less invasively, and less expensively. It has also created uncertainties for laboratories: What is the value of a higher detection rate versus a low false-positive rate? Is the standard of care keeping pace with widespread practices? Should laboratories maintain a first-trimester screening program, a second-trimester program, or both? What algorithms produce the best measures of risk? And are there ever likely to be markers that can keep uncertainty to an absolute minimum?

Counter-trends

Since the first prenatal diagnosis through amniocentesis was made in 1968, the rate of Down births has declined considerably—by about 50 percent. But over the last two decades, the average maternal age has risen by one to three years. "As a result, the number of Down syndrome fetuses being created has gone up," says Andrew R. MacRae, PhD, director and CEO of The Research Institute at Lakeridge Health, Oshawa, Ontario, Canada.

"Against that is the increasing strength of the screens," Dr. MacRae adds. "But if the woman isn’t screened, then there will be more Down cases. If half of all pregnancies go unscreened, already you’re not going to detect half of the cases." While about 65 percent of women are screened in the United States and around 70 percent in the United Kingdom, "we estimate only seven percent of women are screened worldwide," says Yvonne Parker, global prenatal business manager for PerkinElmer Life Sciences, which has its U.S. headquarters in Boston.

Laboratories today can assess the risk of Down syndrome far more accurately than they could in the 1970s, when the only screening marker was maternal age. "Simply asking a patient’s age is a way of assigning risk," Dr. MacRae notes. "Many patients and health care providers alike don’t recognize that maternal age is a screening test, and a poor one at that. From 1970 to 1985, maternal age was the only screening process, and if you were deemed at high risk, i.e. over 35, then the followup test was amniocentesis. Age alone detected about 30 percent, but for that detection, 10 percent of the population had to have amniocentesis." Those patients had a one in 200 risk of procedural-related loss of pregnancy.

With the addition of alpha-fetoprotein, between 1985 and 1990 the detection rate rose to between 35 and 40 percent, while 10 percent still screened positive and had to undergo amniocentesis. But since 1990, U.S. laboratories have increasingly opted to provide the triple screen in the second trimester: alpha-fetoprotein, unconjugated estriol, and human chorionic gonadotropin hormone. High levels of hCG combined with low levels of MSAFP or estriol, or both, have been associated with increased risk. Once the triple screen came into use, the detection rate doubled to 70 or 75 percent, and the number of amniocenteses dropped to seven or eight percent.

Recently, a fourth marker emerged as an adjunct to the triple screen: dimeric inhibin A, or DIA. Produced by the ovaries and the placenta, DIA levels increase during the first trimester until 10 weeks of gestation, remain stable until about 25 weeks, then rise to peak levels at term. But in Down pregnancies, several studies show DIA levels are about twice as high as in unaffected pregnancies.

"It was postulated for years that this substance must be in the body," says Dr. MacRae. "But it’s only been found and measured accurately in the last 10 years." When combined with other serum marker results, DIA information can help identify 75 to 80 percent of Down pregnancies with less than a five percent false-positive rate.

Attempts to use some of the same markers earlier in pregnancy have been disappointing. In one study, only 17 percent of Down syndrome cases were detected by first-trimester AFP testing, and only four percent by first-trimester unconjugated estriol tests.

But the prospects for earlier testing brightened in 1998, with a study by the Foundation for Blood Research, Scarborough, Me. The study showed that by using an hCG test or a free beta subunit hCG test with a test for pregnancy-associated plasma protein A in a multiple marker protocol in the first trimester, researchers could predict more than 60 percent of women carrying Down syndrome fetuses. The study supported earlier research, "which indicated that first-trimester biochemistry could detect about 60 percent of Down syndrome cases," says mathematician David Krantz, director of biostatistics for NTD Laboratories, Huntington Station, NY.

Such assays are not yet licensed for clinical use in the United States, but interest in earlier testing is high.

First-trimester debate

First-trimester serum screening is not yet common in Europe, Canada, or Australia, but it is considered to be the future there, says PerkinElmer’s Parker. Interest in the U.S. is lower. "We don’t do first-trimester maternal serum screening at this time," says Geraldine McDowell, PhD, director of biochemical genetics for Laboratory Corporation of America, Research Triangle Park, NC. "For a variety of reasons it hasn’t quite caught on in the U.S. I believe only a couple of laboratories currently offer it as a routine clinical service; most do it only as part of a research protocol." She expects, however, that this will change in the future. "I’m hearing more and more interest in first-trimester screening among doctors out there, and there are several large, multi-center studies going on in the U.S. right now that are looking at first-trimester screening."

One reason it hasn’t caught on is that the American College of Obstetricians and Gynecologists has recommended against routine clinical use of first-trimester screening. ACOG’s main objection, Dr. McDowell says, is that first-trimester screening usually requires the measurement of nuchal translucency—the amount of fluid accumulation behind the neck of the fetus—via ultrasound, in addition to the serum tests, and the accuracy of ultrasound analysis can vary widely. "ACOG feels it’s not being done in a really standardized way throughout the U.S.," she says.

According to an ACOG report, studies involving low-risk women in a routine clinical practice setting have produced conflicting results that appear to stem from interobserver or intraobserver variation in nuchal translucency measurements. "In addition," the report says, "there is no consensus on the definition of increased nuchal translucency." Alone or in combination with serum markers, nuchal translucency "appears promising but remains investigational."

"It’s analogous," Dr. MacRae says, "to having 20 different laboratory instruments recording your AFP results. You’re trying to keep them in harmony so the results are the same, but given the many different ultrasonographers, their different instruments and levels of experience and skill, you cannot simply assume all those nuchal translucency measurements can be combined and used as consistent pieces of information about the risk of Down’s."

"Nuchal translucency is a physical finding as opposed to a biochemical finding and could be associated with a diverse set of circumstances putting the fetus at risk," Dr. MacRae adds. "Down syndrome is the most common. By itself or with maternal age, nuchal translucency is probably the best single marker for Down’s, but it’s still not sufficiently good to operate on its own."

Combination screening on the rise

NTD Laboratories believes that combination screening is an idea whose time has come. The company’s Ultrascreen, launched about a year ago, combines an ultrasound exam to measure nuchal translucency with serum tests for free beta hCG and pregnancy-associated plasma protein A, or PAPP-A.

Krantz reports that in a study the company published last year, the test proved highly predictive. Studying 10,000 patients between September 1995 and June 1998, the researchers found that by using nuchal translucency and biochemistry in the first trimester, they could detect 90 percent of the Down syndrome cases with a five percent false-positive rate.

"That’s an improvement of about 20 percent on traditional triple tests" done in the second trimester, he notes. "There are also inherent advantages to doing first-trimester screening. The patients are able to learn whether there’s anything wrong with the baby, perhaps before others even know they’re pregnant." Sometimes in the second trimester, he adds, after screening and a followup test, the pregnancy may be close to 20 weeks or even later, which can leave the patient with few options.

NTD continues to collect outcome data and is involved with a related multi-center study funded by the National Institutes of Health, known as the EATABUN trial. For NTD’s purposes, Krantz says, the most important research questions are how to more precisely define detection efficiency, which tends to fluctuate from study to study, and what the false-positive rate is for a larger number of samples.

In answer to critics who say nuchal translucency is prone to measurement error, Krantz agrees that measuring NT requires training: "A physician can’t just decide to do NT." In the NIH-funded study, which is to be presented at the plenary session of the annual meeting of the Society for Maternal-Fetal Medicine in January, "for each center that participated there was extensive training beforehand, and we looked at how people performed as time went on," he says. "There were some questions, when this first began, whether it would be feasible in the U.S. to offer the NT exam, because of the historically wide variation in performance of ultrasound." But he says most of those questions have been answered.

Europe has stayed ahead of the United States on many prenatal screening measures, Krantz says, including the ultrasound measure of nuchal translucency, which was developed by professor Kypros Nicolaides in London, and use of chorionic-villus sampling in the first trimester, which until recently was more often used in Europe than in the United States.

Comparing predictive ability

Jacob Canick, PhD, is principal investigator for the laboratory aspect of the FASTER trial, the NIH-funded first- and second-trimester evaluation of risk of aneuploidy study. Dr. Canick says the study will compare the combination test with the best available second-trimester test, which is a pure serum test: triple markers plus inhibin. "The unique part of the FASTER trial is that all 36,000 enrolled asked to have both tests. They were not randomized. For statistical purposes, that’s the most powerful way to compare the quality of the first-trimester test against the second-trimester test," says Dr. Canick, director of prenatal and special testing for Women and Infants Hospital of Rhode Island and professor in the Department of Pathology and Laboratory Medicine at Brown University, Providence.

Women who enrolled in the study agreed to have an ultrasound exam for estimation of gestational dating, and measurement of nuchal translucency. They also agreed to contribute a blood sample between 10 and 13 weeks. "Then she comes back between 15 and 18 weeks to have another blood sample drawn for the quadruple test," he says.

Another important part of the trial, Dr. Canick says, is that women are not informed of their first-trimester test results until the second-trimester test result is available, at which point they receive both sets of results. "There’s no protocol for offering a first-trimester diagnostic test, which is chorionic-villus sampling," he says. "Enrollees get both screening tests, and if they are positive for either or both, they’re offered amniocentesis in the second trimester."

Much of the controversy about NT stems from the fact that one person at one center has been responsible for so much research on it. "Professor Nicolaides has personally trained lots of people worldwide. He offers a two-day training course mostly in London, but he has satellite courses in the U.S. and Europe and probably around the world at this point," Dr. Canick says. "One purpose of the FASTER trial was to prove whether training in NT can be done properly, independent of professor Nicolaides’ own education process."

The addition of a fourth serum marker has greatly increased the predictive ability of the second-trimester screen, LabCorp’s Dr. McDowell says. Since the triple screen was introduced, its precision has improved, "but the really significant improvement has come from the addition of new markers like inhibin." The quadruple screen that includes inhibin is not in widespread use; its volume is only about 14 percent of that for LabCorp’s triple screen, and the addition of inhibin adds about 50 percent to the cost of the test.

But use is growing every month. "We thought it would be mostly in urban areas—for example, that universities and big medical schools would be the ones jumping onboard, but it doesn’t seem that way. It has more to do with physician awareness. With all the big labs having sales forces out there educating doctors, you don’t necessarily have to be associated with a university setting. And people are probably using it more where the carriers are reimbursing for it."

Market forces

As a sign of what lies ahead, Robert Maciel Associates, which sells its prenatal interpretive software to 25 countries and more than a dozen state programs in the United States, offers inhibin as a third or fourth marker on request, and in the next version of its program will include the ability to screen from weeks 10 to 13 with PAPP-A, hCG or free beta hCG, and nuchal translucency.

"We always have to try to keep up as best we can," says William B. Adams, PhD, president of the Concord, Mass., company. Dr. Adams recalls that there was a huge boom in his business from 1988 to 1993, after the triple screen was introduced, followed by slow, steady growth. Inhibin was somewhat slow to come on to the U.S. market, partly because of the limited availability of test kits. As for first-trimester screening, "I think more will come onboard with that; our software doesn’t yet have first-trimester screening but it will soon, and I think we’ll be in time to catch that," he says.

PerkinElmer Life Sciences has developed a lab information system program called Specimen Gate, an early version of which is used by the state of California to manage data in its unusual public-private partnership to perform prenatal screening. "We contract with eight laboratories across the state to perform both newborn and prenatal testing at 15 to 18 weeks’ gestation," says John E. Sherwin, PhD, chief of California’s Genetic Disease Laboratory. The program handles all prenatal screening performed in the state—about 370,000 women annually.

The state legislature passed a bill in 1986 mandating control of prenatal screening laboratories to ensure not only the quality of laboratory results but also the quality of information on risk calculation, Dr. Sherwin explains. His laboratory establishes methods, provides equipment, manages contracts, and maintains quality assurance. A separate epidemiology group maintains demographic information and algorithms for calculating the medians for various ethnic subgroups.

Although PerkinElmer has set up its prenatal screening software in several European countries, California was the company’s largest project. "One of our goals was to pull together data on a population basis," Parker says. "We’re very much interested in screening from a population point of view. With prenatal screening, the more data you can pull together, the better use you can make of ethnic-specific medians."

Integrated screening debate

Integrated screening, though it requires testing in both first and second trimesters, is a new approach to prenatal screening that recently received a great deal of publicity. Developed by British physician Nicholas Wald, MD, the method combines first-trimester measures of PAPP-A and nuchal translucency with the second-trimester quadruple markers. Using a risk algorithm on these numbers, Dr. Wald has found a detection rate of 85 percent with a false-positive rate of 0.9 percent. "The integrated test detects more cases of Down syndrome with a much lower false positive rate than the best currently available test," he concludes.

An advantage of this approach, Dr. Canick says, is flexibility. "You could actually do just a serum integrated approach, using PAPP-A in the first plus the quadruple test in the second. Then you’d have a single risk estimate on five markers. For those who don’t have nuchal translucency measurement available to them, it would allow for the best pure serum marker test."

Given the difficulty of making proper nuchal translucency certification available to women in outlying areas in the United States, such a test might be a practical alternative for a population-based or public health test. "By adding the first-trimester PAPP-A, you would still be attaining a very high detection rate, but fewer women would need an amniocentesis to achieve that level of detection," Dr. Canick says.

But some critics, particularly in the United States, argue that women are not likely to accept a protocol that maximizes the results by delaying the diagnosis. Despite the increased sensitivity and specificity of integrated screening, they believe many women who get a positive result in the first trimester would be unwilling to wait for additional data in the second trimester. More explicitly, guest contributors to the New England Journal of Medicine (1999;341[7]:521-522) wrote in a commentary on Dr. Wald’s study: "We would find it clinically and ethically troubling not to tell a pregnant woman about a heightened risk of Down’s syndrome at a time when she could have chorionic-villus sampling and undergo elective termination of the pregnancy in a discrete and safe way, should the results be abnormal."

Dr. Wald responds that early testing should not be an end in itself. "All other things being equal, offering a screening test for Down’s syndrome early in pregnancy is better than offering one later. But all things are not equal. The integrated test is considerably better than first-trimester testing alone."

Refining hCG markers

Another recently developed test is also showing great promise: hyperglycosylated hCG (HhCG). Currently being tested by Yale University researchers in a national clinical trial, the test for urine HhCG was developed by biochemist Laurence Cole, PhD, and the patent leased to Quest Diagnostics.

Dr. Cole, professor of obstetrics and gynecology and of biochemistry and molecular biology, and chief of the new Division of Women’s Health Research at the University of New Mexico, Albuquerque, specializes in hCG. It is at the core of the triple and quadruple tests, he notes, and is the most powerful indicator of Down syndrome. On its own, hCG can pick up about 40 percent of cases in the second trimester, and adding AFP and estriol bumps the rate up to about 60 percent, with a five percent false-positive rate. But 40 percent are still missed, so "there’s been very great interest in finding better technology for screening those under 35, those not at high risk," Dr. Cole says.

Since inhibin edges the detection rate up to 70 percent, adding it to the triple screen as a fourth test is a practice that is spreading quickly nationwide, he says. "So now, four markers are thrown into the pot, and this test is pretty much replacing the triple screen. But people still aren’t happy, because an awful lot of them have to have amniocentesis to identify that tiny percentage with Down’s."

The focus on first-trimester screening, he says, stems from the fact that if you can screen then, and have to terminate the pregnancy, you may be terminating a three- or four-month pregnancy. But first-trimester screening is not as good because roughly half of the Down cases present then would be miscarried by the second trimester. "When you’ve gotten to the second trimester, you’ve eliminated roughly half. So if you have a 50 percent detection rate in the first trimester, it’s nothing like as good as a 50 percent detection rate in the second."

In Dr. Cole’s research, "We noticed about five years ago that the hCG produced by women with a proven Down syndrome pregnancy is a fraction less stable than that in a normal pregnancy; it breaks up into parts or alpha-beta subunits. That led us to suspect there was a structural difference in the hCG produced in a Down’s pregnancy," he says.

"We had developed an antibody jointly with Columbia University that recognized hCG with bigger sugar side chains, so we immediately wanted to see if this antibody could distinguish Down’s from non-Down’s," Dr. Cole continues. His group of researchers did this by collecting 1,500 urine samples from women undergoing amniocentesis.

Of those, 39 had genetic or karyotype-proven trisomy 21, and after the data were sent to a statistician, the hyperglycosylated hCG test proved to have an 80 percent detection rate and five percent false-positive rate, and a rate of 95 percent when it was combined with AFP levels. Quest, which is conducting large clinical trials of the test in California, has named the test invasive trophoblast antigen, or ITA.

Urine tests have problems, Dr. Cole admits, because almost all machines are designed for blood, and urine has to be normalized to some dilution factor. But he is optimistic about the potential for HhCG to serve as the primary screening marker. "For women under 35, I’m hoping that the final formulation will end up at 90 percent detection and three percent false positive, or maybe even come down to one percent or two percent." However, he says, the research is going slowly, and "there’s nothing to stop something better from appearing."

Subdued FDA role

Perhaps uncharacteristically, the Food and Drug Administration has been anything but heavy-handed about regulating prenatal screening tests. Historically, maternal screening began with testing alpha-fetoprotein levels in the 1970s, when it was discovered that elevated levels were a risk factor for anencephaly and spina bifida. In the early 1980s, Dr. Canick recounts, the FDA chose to regulate the licensing of AFP diagnostic kits for neural-tube defect screening, so in 1983 the agency allowed kits to be released with the claim that they were neural-tube defect tests.

"Then in 1984, when AFP was found to be a marker for Down syndrome pregnancy, which was an unexpected finding, it was very easy for people already doing AFP screening to add that use. The FDA kept away from that. In the late 1980s, when new markers started being shown, the FDA considered allowing companies to submit data for validation and FDA approval, but it never went anywhere. I would say the FDA has chosen to have a hands-off policy on screening for Down’s."

Says Dr. McDowell: "All of the kits we use are FDA-approved. However, none of them is approved for screening for Down and trisomy 18, even though screening for that has been around for a couple of decades and it’s considered the standard of care. So we do have a disclaimer on our reports, saying that uses as Down’s and trisomy 18 screeners are investigational. Really, in practice they’re not, and insurance companies reimburse for them."

She suggests that a large and expensive study would be required to demonstrate the clinical validity of the screening markers for Down—and the payoff would be small. "I don’t see that any companies would gain any markets. Everyone is already using the test and getting reimbursed."

The FDA’s official policy primarily affects manufacturers of test kits, says Dr. Adams of the software firm Robert Maciel Associates. "We sell only customized software to each laboratory, and as customized software we fall under the radar of the FDA. But no manufacturers can mention Down’s syndrome in their test kits." In many ways, he suggests, it would be a good idea for the FDA to regulate software. "But I would be afraid for us as a small company. It might be a huge financial burden to go through the FDA’s hoops."

Strong overseas prospects

"With the FDA situation, the U.S. is very interesting," PerkinElmer’s Parker says. "There is no diagnostic company that is offering software as well, because by doing so they might be making an implied claim on the reagents." By contrast, Europe’s regulatory environment encourages the manufacturer to provide all components; under the French regulatory system, any company supplying prenatal reagents and instrumentation must provide software.

So while PerkinElmer has a significant market share in most European countries, it does not offer its LifeCycle prenatal screening software in the U.S. market. Although California uses the company’s Specimen Gate program, it performs its own risk calculations. "We have not been comfortable with supplying prenatal risk calculation engines because it’s such a gray area," Parker says.

Two-thirds to three-fourths of Robert Maciel Associates’ new sales are overseas, says Dr. Adams, noting that his company doesn’t advertise in the United States. Although a handful of U.S. laboratories need interpretive software because they have decided to bring their testing in-house, few new labs are picking up prenatal testing. Moreover, "labs have no inclination to change their software," he says. "It’s a fairly wrenching process. Frequently they may have years worth of data stored under one system, and when they go to another, it may or may not translate."

Dr. Adams’ company entered the European market in 1992 and became an approved prenatal software provider for the French government in 1997 as part of an arrangement with Abbott Laboratories. Taiwan, China, Korea, India, Israel, Turkey, and four countries in Africa have also purchased the software. "Our latest is New Caledonia, a tiny strip of island north of New Zealand. They found out about us through Abbott France," Dr. Adams says.

At laboratories’ discretion

In the United States, where prenatal screening practices are far less centralized, individual laboratories must make sometimes difficult decisions. For example, laboratories use different AFP cutoffs for neural-tube defect screening. "That’s one place where not all laboratories do the same thing," Dr. McDowell says.

A recent CAP Survey (FP-A 2000) found that approximately one-third of labs use 2.0 multiples of the median, and about half use 2.5 multiples of the median. The lower MoM cutoff will detect about 85 percent of open spina bifida while the higher cutoff will detect about 70 percent of spina bifida. But the tradeoff is that the 2.0 cutoff will have approximately twice as many screen positives as the 2.5 MoM cutoff—four percent versus two percent.

"So you have to say, well, to catch those extra 15 percent of affected fetuses, is it worth referring those additional women for amniocentesis? Sending someone for amniocentesis involves a risk of miscarriage and considerable anxiety. Is that properly offset by the higher detection rate? This is a decision that labs have to make, and different labs have made different decisions," Dr. McDowell says.

For Down syndrome screening, most laboratories use a second-trimester risk cutoff of one in 270, she says. "With the triple test, this risk cutoff will detect 60 percent of cases of Down syndrome, with five to eight percent of all women having a screen-positive result. Women with positive screens require followup procedures such as ultrasound examination and/or amniocentesis. That’s what you generally get with that one in 270 risk cutoff," she says. "So you know you’ll pick up about 60 percent of the pregnancies that are affected with Down’s. So 40 percent of women who have a child with Down’s will screen negative. The flip side is that five to eight percent of the people you’re testing will come back screen-positive but, in fact, the majority of them are going to be false positives.

"Sometimes we will also see things that indicate other abnormalities, not necessarily what we were screening for," Dr. McDowell adds. "For example, there are a couple of different genetic disorders that will cause one of the analytes in the panel, unconjugated estriol, to be undetectable. The most common cause is steroid sulfatase deficiency, which causes ichthyosis, a skin abnormality. Isolated STSdeficiency is relatively benign, but a more rare form of STS deficiency is caused by a large chromosome deletion that knocks out other genes in addition to steroid sulfatase. This contiguous gene syndrome is clinically more serious than isolated STS deficiency. In cases such as these, our laboratory will have a genetic counselor contact the physician to discuss the implications and appropriate followup."

Other markers, with unknown implications for the laboratory, appear to be just around the corner. A paper by Nicolaides, et al, published recently in The Lancet (2001;358 [9294]:1665-1667), reports on another first-trimester test, the nasal bone ultrasound marker. Although the study has not yet been replicated, "The detection rate and false-positive rate achievable with this approach are particularly impressive and may negate the requirement for any serum testing," Parker says.

But if the test is adopted, it could be five or 10 years down the road before it becomes routine practice, and she does not believe the approach will obviate the need for serum testing. "Personally, I think it will make the case very strongly for first-trimester screening and biochemistry will still be required," says Parker. "Ultrasound markers on scans require skilled experts to interpret, and for that reason, I believe that biochemistry will be around for some years yet."

Anne Paxton is a writer in Seattle. For more information about Down syndrome risk, visit the CAP Web site.

   
 

 

 

   
 
 © 2014 College of American Pathologists. All rights reserved. | Terms and Conditions | CAP ConnectFollow Us on FacebookFollow Us on LinkedInFollow Us on TwitterFollow Us on YouTubeFollow Us on FlickrSubscribe to a CAP RSS Feed