junct to help detect structural abnormalities. In some countries, ultrasonography is performed on every pregnant woman, and its use is increasingly common in the United States as well. It may be the most frequently used prenatal diagnostic test. The general consensus is that ultrasound should not be used routinely in every pregnancy (NIH, 1984; ACOG, 1988), but that consensus may not reflect actual obstetrical practice today. Recent reviews in the literature have identified some factors affecting reliability in identification of fetal anomalies through ultrasound (Filly et al., 1987; Goldstein et al., 1989); others have reported variability in quality by center and practitioner and error rates of more than 10 percent (Levi et al., 1991; Lancet, 1992). The committee heard testimony that raised its concern about the reliability of interpretation of prenatal ultrasound images and about the consequences of decisions based on such interpretation (see Chapter 2).
Genetic tests are seldom perfect predictors of clinical risk. No biochemical screening test is sensitive enough to detect all cases, and even current DNA methods cannot detect all possible mutations that cause a specific disease. For example, more than 300 mutations in the CF transmembrane regulator gene have been found that cause CF; most of them are extremely rare. Current technology permits the simultaneous detection of approximately 20 of the most common ones, accounting for about 85 to 90 percent of mutations in whites but a much smaller proportion of mutations in nonwhites (Cutting et al., 1992). Moreover, the frequency of the different mutations for CF can vary within subpopulations; what applies to a northern European population does not necessarily apply to Italian or Jewish populations. The 0.4 to 0.6 percent of normal individuals who carry undetectable mutations will not know—even after testing—that they are at risk, if they mate with another carrier, of having a child with CF. Similar allelic diversity occurs in many other disorders. Thus it becomes important to distinguish between analytic sensitivity, the ability of a test to detect the various mutations it was designed to detect, and clinical sensitivity, the ability of the test to detect all patients who will get, or who have, the disease.
A way could be found around the less-than-perfect sensitivity of current DNA tests in the presence of allelic diversity. The multiple mutations that result in a specific disease all impair the normal expression of the gene. Consequently, a test that measures normal gene expression could tell, indirectly, whether any mutation capable of altering expression is present. Tests of altered gene product antedated DNA-based tests, but they could only be performed when evidence of gene expression could be found in readily accessible tissues. Recent work indicates that minute amounts of mRNA for proteins that are detectable only in specialized tissues are present in accessible tissues such as white blood cells and cultured chorionic villus cells (Chelly et al., 1989; Sarkar and Commer, 1989). The protein could be translated from the mRNA and its structure and function assayed.