I have been asked to write a commentary on the article by Andaloro et al. (1), the letters-to-the-editor by Drs. Polifka and Shepard and by Dr. Brent, and the response to those letters (2–4) by Dr. Rosenquist, the senior author of the original article. This article and these letters raise extremely important questions about the best way to delineate exposures in pregnancy that are harmful to the fetus i.e. human teratogens. I appreciate having this opportunity to put this exchange of strong differences of opinion into a broader perspective.

Everyone agrees on the importance of an early identification of harmful fetal exposures so that these can be prevented whenever possible. While the readers of Pediatric Research or any other journal agree on this concern, I would predict that most will not feel well-prepared to evaluate this reports that proclaim the identification of a new human teratogen. Nor should they. Instruction in the clinical aspects of harmful fetal exposures, such as the methods for delineating these or the characteristics of a human teratogen, is not taught in most medical schools or residency programs. Most doctors do not know the name of a reliable reference book to use (5–7) or on-line resources, such as REPROTOX and TERIS, that are available through many hospital libraries. Even more disappointing, most physicians do not realize that the information about pregnancy risks in the Physicians Desk Reference (PDR) is not accurate. They have not heard about either the scientific assessments (8) or the professional organizations, such as the Teratology Society (9), which have concluded that the Pregnancy Classification for drugs i.e. classes A, B, C, D, and X, is not accurate as used and should be eliminated. Efforts are just beginning to encourage the development of prospective pregnancy registries, which will provide better information on the potential risks from exposures than is now available.

The characteristics to be looked for in identifying a human teratogen have been delineated by several experienced teratologists (10–12). The major features are:

  1. 1

    There is an increased frequency of a specific phenotype in comparison to that in an unexposed population.

  2. 2

    The frequency of the adverse effect shows a dose-response relationship.

  3. 3

    There is an animal model.

  4. 4

    There is a genetically more susceptible group of exposed individuals.

  5. 5

    The proposed teratogenicity is plausible biologically.

With these characteristics in mind, how is the physician-scientist reader to interpret the findings reported by Andaloro et al.? Equally important, how were the reporters to know how to evaluate the information that was presented at the press conference held to announce these findings? This discussion starts with the central question: can human teratogens be identified by the in ovo studies used by Andaloro and his associates. The reasons why an in ovo or any type of in vitro study cannot be used to identify a human teratogen have been outlined in detail by Drs. Polifka and Shepard and Dr. Brent in their letters-to-the-editor (2–4). The major limitations are: 1) the absence of the normal physiologic process of intestinal absorption and metabolism of the compound by the mother; 2) the dose of dextromethorphan used, which was much higher than the human therapeutic dose on a mg/kg basis; 3) the fact that the experimental model did not identify affected fetuses at the end of gestation, but killed the embryos after 72 h; 4) the absence of any convincing data of a human risk from the epidemiologic studies available (13,14). This article by Andaloro et al. (1) did not raise the issue for the first time; the comparison of the findings in in vitro and in vivo studies have been the subject of many expert panels (15) and review articles (16,17). To date, it has not been possible to use in vitro studies to identify human teratogens or to predict the findings in in vivo studies. However, in vitro studies clearly have value as a technique for examining cellular and molecular mechanisms of the action of compounds and structure-activity relationships.

As a clinician who counsels pregnant women and their doctors about the potential danger of exposures during pregnancy, I was most concerned to see the headlines that developed from the press conference held to announce the findings by Andaloro et al. (1). They included: "Dextromethorphan Linked to Birth Defects" and "Cough Syrups Hurt Embryos." Not only is this an inappropriate extrapolation, but it could lead to inappropriate decisions to terminate an exposed pregnancy because of the anxiety produced by this announcement. We should not be surprised at how the reporters interpreted what was said; they also have no training in teratology. To put this into perspective, we should remember the consequences of the publicity about the nationwide ban in 1973 on spray adhesives by the U.S. Consumer Product Safety Commission because of a report of an association of chromosome breakage and birth defects. Hook and Healy showed (18) that the genetic centers they polled reported having received 1100 inquiries from concerned individuals about the potential risks in pregnancy. Of those who inquired, 11 women chose to have an amniocentesis to assess the potential risk of chromosome breakage and 9 women had an elective termination of pregnancy.

Each year several exposures in pregnancy are postulated to be teratogenic. Sometimes the hypothesis is carried in an abstract at a scientific meeting, sometimes in a journal article and sometimes in a television news program. Clinicians need an ability to assess these hypotheses quickly. When Adaloro et al. (1) published their article, only two sources of information were available on the findings in human pregnancies: 1) the data from the National Collaborative Perinatal Project, a study of 50 282 mother-child pairs whose pregnancies lasted at least 5 months in the years 1959-1965 (13); 2) the findings in the liveborn infants of 6 509 mothers in an HMO in 1980-1982; a correlation had been made between automated pharmacy records and the outcome of the pregnancy recorded in the medical record (14). One would have preferred more data and from studies with different designs.

How can the situation be improved? Theoretically the clinician concerned about a postulated teratogen could call the manufacturer of the drug of concern and ask the scientific staff for information about the findings in the animal studies of dextromethorphan that were required (19) before marketing. I tried that in the case of dextromethorphan with several products and manufacturers and tired of the endless referral to other individuals and answering machines. This option would only work if the clinician knew in advance the name and telephone number of a company scientist who was authorized to (and willing to) respond to such inquiries.

Two new developments may make a difference in making more current information on common exposures available, rather than having to wait for the results of a study stimulated by a new theory and performed over a few years' time. One is the development by manufacturers of pregnancy registries in which women who have used their product will be enrolled prospectively, i.e. before any prenatal screening has been done. Registries have been established at the manufacturer's facilities for antiviral agents, e.g. acyclovir (GlaxoWellcome), and the vaccine for varicella (Merck), as well as at hospitals, e.g. the antiepileptic drug registry, which we administer with support from six companies. Another theoretical possibility would be to develop data from the new National Birth Defects Prevention Study, in which eight statewide birth defects surveillance programs are coordinated through the Centers for Disease Control (CDC) in Atlanta. In this study each state will enroll and interview each year 300 mothers of infants with major malformations and 100 mothers of normal infants. The mother of each enrolled infant is interviewed by telephone about exposures in pregnancy, as well as other factors, such as demographic characteristics, medication taken, diet, and occupation. Theoretically the 3200 annual interviews (2400 malformed and 800 nonmal-formed) will develop a valuable new database that could be analyzed with regard to common over-the-counter drugs, such as dextromethorphan. Whether this information will be developed and available for clinicians has not been determined. However, this resource could address many questions concerning exposures in pregnancy that will continue to arise.

One would prefer that the readers of Pediatric Research become more interested in the delineation of human teratogens out of intellectual curiosity. If the strong difference of opinion concerning the interpretation of the article by Andaloro et al. increase the level of interest in this topic as a side benefit, that is still progress.

Lewis B. Holmes, M.D.

Genetics and Teratology Unit; Warren 801; Massachusetts General Hospital; 55 Fruit Street Boston, MA 02114