- Email: [email protected]
Potential prenatal predictions of Down syndrome: A statistical analysis
December 1990 Am J Obstet Gynecol
Indik and Reed
venous and arterial velocities must be considered in the interpretation of the changes in umbilical arterial systolic and diastolic velocities reported in fetuses with growth retardation.
5.
REFERENCES 1. Trudinger BJ, Giles WB, Cook CM. Flow velocity waveforms in the maternal uteroplacental and fetal umbilical placental circulation. AM J OBSTET GYNECOL 1985; 152: 15563. 2. Schulman H. The clinical implications of Doppler ultrasound analysis of the uterine and umbilical arteries. AM J OBSTET GYNECOL 1987; 156:889-93. 3. Trudinger BJ. The umbilical circulation. Semin Perinatal 1987;11:311-21. 4. Divon MY, Guidetti DA. Doppler ultrasound-a simple
6.
7. 8.
technique for identification of respiratory sinus arrhythmia (RSA) in the human fetus. In: Proceedings of the 34th Annual Meeting of the Society for Gynecologic Investigation, March 18-21, 1987. Atlanta, Georgia: Society for Gynecologic Investigation, 1987 :215. Hoskins PR, McDicken WN,Johnstane FD, White G, Haddad NG, Chambers SE. Determination of the presence of fetal apnea using umbilical artery and umbilical vein Doppler waveforms. Ultrasound Med BioI 1988; 14:589-92. Reed KL, Anderson CF, Shenker L. Changes in intracardiac Doppler blood flow velocities in fetuses with absent umbilical artery diastolic flow. AM J OBSTET GYNECOL 1987;157:774-9. Reed KL, Sahn DJ, Marx GR, Anderson CF, Shenker L. Cardiac Doppler flows during fetal arrhythmias: physiologic consequences. Obstet GynecoI1987;70:1-6. Reuss ML, Rudolph AM, Dae MW. Phasic blood flow patterns in the superior and inferior venae cavae and umbilical vein of fetal sheep. AM J OBSTET GYNECOL 1983; 145:70-8.
Potential prenatal predictions of Down syndrome: A statistical analysis William F. O'Brien, MD, Robert A. Knuppel, MD, MPH, Carol Torres, BS, and Deborah Sternlicht, RN, MS Tampa, Florida To determine the feasibility of combining several screening tests for the prenatal detection of Down syndrome, we evaluated the potential relationship among three proposed predictors. We determined the concentration of chorionic gonadotropin in frozen serum samples from women of known maternal age and weight, fetal biparietal diameter, and femur length, and a-fetoprotein concentration. When corrected for gestational age and maternal weight, the potential predictors were independent, except for a slight correlation (r = 0.10) between maternal serum a-fetoprotein and maternal serum human chorionic gonadotropin. Both maternal serum human chorionic gonadotropin and biparietal diameter/femur length demonstrated an approximately log-normal distribution similar to maternal serum a-fetoprotein. Therefore it is scientifically sound to use any or all of these variables in combination for the identification of pregnancies at increased risk for Down syndrome. (AM J OSSTET GYNECOL 1990;163:1796-8.)
Key words: Prenatal diagnosis, Down syndrome, a-fetoprotein Because recognition of the association of fetal Down syndrome with a low maternal serum a-fetoprotein (AFP), a number of investigators have demonstrated the feasibility of the combined use of maternal age and maternal serum AFP concentration for the selection of women at increased risk for Down syndrome. In addition to maternal serum AFP, several potential mark-
From the Department of Obstetrics and Gynecology, University of South Florida College of Medicine. Presented at the Thirty-seventh Annual Meeting of the Society for Gynecologic Investigation, St. Louis, Missouri, March 21-24, 1990. Reprint requests: William F. O'Brien, MD, Department of Obstetrics and Gynecology, University of South Florida, College of Medicine, Suite 500, 4 Columbia Dr., Tampa, FL 33606.
6/6/24703
1796
ers for Down syndrome have been proposed. Of these, the concentration of human chorionic gonadotropin (hCG) in maternal serum' and the fetal biparietal diameter/femur length rati0 2 seem to be the most promising. Because the bayesian analysis of risk requires that assignment of final risk can only be achieved if the indicators of relative risk are independent, we sought to determine whether these proposed markers were independent of each other and of maternal age, and to examine the distribution of values in normal pregnancies.
Material and methods Specimens were obtained from frozen samples stored at - 70° C. These samples had been submitted to the
Potential predictors of Down syndrome
Volume 163 Kumber 6, Part I
University of South Florida Regional Screening Program from Jan. I, 1987, through Dec. 31, 1988. The following criteria were required for choice of samples: ultrasonographic measurement of biparietal diameter and femur length performed on the day of serum sampling, menstrual dating agreeing within 10 days of ultrasonographic dating, and gestational age between 15 and 19 completed weeks. When these criteria were met in more than 200 samples for a given gestational week, every other sample was chosen. Maternal age and ultrasonographic measurements, as well as assurance that the infant was phenotypically normal, were provided by the referring physician. Both maternal serum AFP and hCG concentrations were measured by enzymelinked immunosorbent assay with commercially available kits (Abbott Laboratories, Abbott Park, Ill.). Determinations were made in duplicate and the coefficient of variation was required to be < 10%. For convenience in comparison, results for maternal serum AFP were expressed as international units per ml and maternal serum hCG was expressed as international units per liter. Samples were grouped according to completed gestational weeks and weekly medians for maternal serum AFP, and maternal serum hCG, and biparietal diameter/femur length ratios were determined. Smoothed medians were then calculated by weighted linear regression. Adjustment of maternal serum AFP and maternal serum hCG for variation in maternal weight was by regression of the nonadjusted multiples of the median with reported maternal weight. All analyses were accomplished with the SAS statistical packages (SAS Institute, Cary, N. C.).
1797
Table I. Smoothed weekly medians for possible screening parameters Maternal
Week
n
serum AFP
15 16 17
76 146 130 120
24.1 29.4 34.7 40.4
18
Maternal
serum keG
BPD/FL
25.0 21.0 17.0 12.7
1.73 1.66 1.60 1.53
BPD/ FL, Fetal biparietal diameter/femur length ratio.
Table II. Relationship between predictors of Down syndrome Variable 1
Maternal Maternal Maternal Maternal Maternal
serum serum serum serum serum
Variable 2
heG AFP heG AFP heG
r
p
-0.04 0.37 BPD/FL BPD/FL 0.06 0.16 0.05 0.22 Age 0.03 0.45 Age Maternal serum AFP 0.10 0.02
BPD/FL, Fetal biparietal diameter/femur length ratio; Age, maternal age; T, Spearman Rank correlation coefficient; p, probability.
the median closely approximated gaussian distribution with a Shapiro-Wilk statistic value of 0.98, whereas maternal serum AFP achieved a value of 0.96 (values may range from 0 to 1.0; 1.0 indicates completely normal distribution). The distribution of biparietal diameter/femur length with a value of 0.88 differed more from a completely normal distribution value of 0.88, but conformed sufficiently to allow clinical usage. Comment
Results
There was no correlation between maternal age and any of the three variables, either as uncorrected values or after transformation. Initial analysis confirmed the expected correlations between gestational age and maternal serum AFP (r = 0.23), maternal serum hCG (r = - 0.25), and biparietal diameterlfemur length ratio (r = - 0.33). Smoothed weekly medians for each variable were generated (Table I) and values were transformed into multiples of the median. This transformation nullified an association with gestational age. However, a significant positive correlation remained between maternal serum hCG and maternal serum AFP (r = 0.15). Because maternal serum AFP (r = - 0.18) and maternal serum hCG (r = - 0.23) were both varied inversely with maternal weight, both were subsequently expressed as weight-adjusted multiples of the median. After this transformation, there was no longer a correlation with maternal weight, but a small (r = 0.10) but statistically significant correlation between maternal serum AFP and maternal serum hCG remained (Table II). Logarithms of maternal serum hCG multiples of
The traditional method of patient selection for Down syndrome testing by amniocentesis has been maternal age. Unfortunately, with use of this criterion alone only 20% to 30% of women with a fetus with Down syndrome are offered amniocentesis.' After the initial report that women carrying fetuses with Down syndrome had lower than average maternal serum AFP values, t this observation was confirmed and tables allowing the use of combined maternal age and maternal serum AFP concentration to estimate the risk of Down syndrome were published. 5 Several recent repons 6 . 7 have confirmed the value of combining maternal age with maternal serum AFP values to select a population of women younger than the traditional cutoff age with a risk similar to that of older women. Unfortunately, even with the a~junctive, use of maternal serum AFP detection of a significantly increased proportion of fetuses with Down syndrome requires a high false-positive rate. s The potential of using a combination of noninvasive measures to increase the sensitivity or decrease the false positivity of Down syndrome screening is of great cur-
1798 O'Brien et al.
rent interest. Of the possible candidates for such a role, the biparietal diameter/femur length ratio and maternal serum hCG appear to be the most promising. Inclusion of the biparietal diameter/femur length ratio is unlikely to result in a significant increase in cost, because most programs require ultrasonographic confirmation of dating before final recommendation for amniocentesis. Unlike maternal serum AFP, which is measured primarily to screen for open neural tube defects, determination of maternal serum hGC would increase screening costs. Currently, however, it appears to be the most likely candidate as a second screen since it performs more efficiently than measurement of unconjugated estriol at similar cost. In this study we demonstrated that, after adjustment for gestational age and maternal weight, maternal serum AFP, maternal serum hCG, and biparietal diameter/femur length ratio are sufficiently independent from each other and from maternal age that they may be combined in a bayesian estimate of relative and final risk. The correlation between maternal serum AFP and maternal serum hCG found in our sample before adjustment for maternal weight (0.15) is similar to the value (0.14) reported by Wald et a1. 9 in samples for which maternal weight was not available. However, maternal weight was only partially responsible for the corE relation and it is probable that there is an underlying biologic association. In contrast, the lack 6f association between biparietal diameter/femur length ratio and either of the biochemical measures assures that the combination of retarded femur growth, increased production of hCG, and deficient production of AFP seen in Down syndrome does not occur in chromosomally normal fetuses. Although these results are encouraging, care is required before the widespread introduction of maternal serum hCG and biparietal diameter / femur length ratio for Down syndrome screening. First, estimation of relative risk on the basis of the distributions of normal and Down syndrome pregnancies is relatively straightforward, but there is some concern that the distribution of maternal serum AFP values in pregnancies associated with Down syndrome may be dependent on maternal age or on geographic site. IO The distribution of maternal serum hCG in pregnancies associated with Down syndrome has only been determined in a small number of cases, and similar concerns must be considered. Second, the values for the biparietal diameter/femur length ratios noted in this study are similar to those reported from Boston, but they differ from those generated in New Haven" with intermediate results reported from Pittsburgh. '2 Although it is possible
December 1990 Am J Obstet Cynecol
that these differences are based on population, it is more likely that they are related to technique. 2 Unlike maternal serum AFP or maternal serum hCG determinations, over which laboratories can exert quality control, most centers rely on ultrasonographic values generated by the referring physician. Significant variations in ultrasonographic technique among participating physicians may obscure the value of biparietal diameter/femur length ratio on screening. Finally, the cost effectiveness and goals of the screening program must be evaluated. Should maternal serum hCG be determined for all subjects or only those found at increased risk by evaluation of maternal age and maternal serum AFP? We have shown that the combination of maternal age, maternal serum AFP, maternal serum hCG, and biparietal diameterlfemur length ratio in the prediction of Down syndrome risk may be feasible. The proper implementation of such a program requires further study. REFERENCES 1. Bogart MH, Pandian MR,]ones OW. Abnormal maternal serum chorionic gonadotropin levels in pregnancies with fetal chromosome abnormalities. Prenat Diagn 1987; 7: 623-30. 2. Lockwood C, Benacerraf B, Krinsky A, et al. A sonographic screening method for Down syndrome. AM] OBSTET GYNECOL 1987;157:803-8. 3. Adams M, Erikson], Layder P, Oakley GP. Down's syndrome: recent trend in the United States. ]AMA 1981; 246:758-60. 4. Merkatz lR, Nitowski HM, Macri ]N, Johnson WE. An association between low maternal serum alphafetoprotein and fetal chromosomal abnormalities. AM] OBSTET GYNECOL 1984; 148:886-91. 5. Cuckle HS, Wald N], Lindenbaum RH. Maternal serum alpha-fetoprotein measurement: a screening test for Down syndrome. Lancet 1984; 1:926-9. 6. DiMaio MS, Baumgarten A, Greenstein RM, Saal HM, Mahoney MJ. Screening for fetal Down's syndrome in pregnancy by measuring maternal serum alphafetoprotein levels. N Engl] Med 1987;317:342-6. 7. Palomaki GE, William], Haddow]E. Combining maternal serum alpha-fetoprotein measurements and age to screen for fetal Down syndrome in pregnant women. AM ] OBSTET GYNECOL 1989;160:575-81. 8. Spencer K, Carpenter P. Screening for Down's syndrome using alpha-fetoprotein: a retrospective study indicating caution. Br Med] 1985;290: 1940-3. 9. Wald N], Cuckle HS, Densem ]W, et al. Maternal serum screening for Down's syndrome in early pregnancy. Br Med] 1988;297:883-7. 10. Hook EB. Variability in predicted rates of Down syndrome associated with elevated maternal serum alphafetoprotein levels in older women. Am] Hum Genet 1988;43:160-4. 11. Lockwood C, Benacerraf B, Krinsky A, et al. A sonographic screening method for Down syndrome. AM] OBSTET GYNECOL 1987; 157:803-8. 12. Hill LM, Guzick D, Belfar HL, Hixson], Rivello D, Rusnak ]. The current role of sonography in the detection of Down syndrome. Obstet Gynecol 1989;74:620-3.
Indik and Reed
venous and arterial velocities must be considered in the interpretation of the changes in umbilical arterial systolic and diastolic velocities reported in fetuses with growth retardation.
5.
REFERENCES 1. Trudinger BJ, Giles WB, Cook CM. Flow velocity waveforms in the maternal uteroplacental and fetal umbilical placental circulation. AM J OBSTET GYNECOL 1985; 152: 15563. 2. Schulman H. The clinical implications of Doppler ultrasound analysis of the uterine and umbilical arteries. AM J OBSTET GYNECOL 1987; 156:889-93. 3. Trudinger BJ. The umbilical circulation. Semin Perinatal 1987;11:311-21. 4. Divon MY, Guidetti DA. Doppler ultrasound-a simple
6.
7. 8.
technique for identification of respiratory sinus arrhythmia (RSA) in the human fetus. In: Proceedings of the 34th Annual Meeting of the Society for Gynecologic Investigation, March 18-21, 1987. Atlanta, Georgia: Society for Gynecologic Investigation, 1987 :215. Hoskins PR, McDicken WN,Johnstane FD, White G, Haddad NG, Chambers SE. Determination of the presence of fetal apnea using umbilical artery and umbilical vein Doppler waveforms. Ultrasound Med BioI 1988; 14:589-92. Reed KL, Anderson CF, Shenker L. Changes in intracardiac Doppler blood flow velocities in fetuses with absent umbilical artery diastolic flow. AM J OBSTET GYNECOL 1987;157:774-9. Reed KL, Sahn DJ, Marx GR, Anderson CF, Shenker L. Cardiac Doppler flows during fetal arrhythmias: physiologic consequences. Obstet GynecoI1987;70:1-6. Reuss ML, Rudolph AM, Dae MW. Phasic blood flow patterns in the superior and inferior venae cavae and umbilical vein of fetal sheep. AM J OBSTET GYNECOL 1983; 145:70-8.
Potential prenatal predictions of Down syndrome: A statistical analysis William F. O'Brien, MD, Robert A. Knuppel, MD, MPH, Carol Torres, BS, and Deborah Sternlicht, RN, MS Tampa, Florida To determine the feasibility of combining several screening tests for the prenatal detection of Down syndrome, we evaluated the potential relationship among three proposed predictors. We determined the concentration of chorionic gonadotropin in frozen serum samples from women of known maternal age and weight, fetal biparietal diameter, and femur length, and a-fetoprotein concentration. When corrected for gestational age and maternal weight, the potential predictors were independent, except for a slight correlation (r = 0.10) between maternal serum a-fetoprotein and maternal serum human chorionic gonadotropin. Both maternal serum human chorionic gonadotropin and biparietal diameter/femur length demonstrated an approximately log-normal distribution similar to maternal serum a-fetoprotein. Therefore it is scientifically sound to use any or all of these variables in combination for the identification of pregnancies at increased risk for Down syndrome. (AM J OSSTET GYNECOL 1990;163:1796-8.)
Key words: Prenatal diagnosis, Down syndrome, a-fetoprotein Because recognition of the association of fetal Down syndrome with a low maternal serum a-fetoprotein (AFP), a number of investigators have demonstrated the feasibility of the combined use of maternal age and maternal serum AFP concentration for the selection of women at increased risk for Down syndrome. In addition to maternal serum AFP, several potential mark-
From the Department of Obstetrics and Gynecology, University of South Florida College of Medicine. Presented at the Thirty-seventh Annual Meeting of the Society for Gynecologic Investigation, St. Louis, Missouri, March 21-24, 1990. Reprint requests: William F. O'Brien, MD, Department of Obstetrics and Gynecology, University of South Florida, College of Medicine, Suite 500, 4 Columbia Dr., Tampa, FL 33606.
6/6/24703
1796
ers for Down syndrome have been proposed. Of these, the concentration of human chorionic gonadotropin (hCG) in maternal serum' and the fetal biparietal diameter/femur length rati0 2 seem to be the most promising. Because the bayesian analysis of risk requires that assignment of final risk can only be achieved if the indicators of relative risk are independent, we sought to determine whether these proposed markers were independent of each other and of maternal age, and to examine the distribution of values in normal pregnancies.
Material and methods Specimens were obtained from frozen samples stored at - 70° C. These samples had been submitted to the
Potential predictors of Down syndrome
Volume 163 Kumber 6, Part I
University of South Florida Regional Screening Program from Jan. I, 1987, through Dec. 31, 1988. The following criteria were required for choice of samples: ultrasonographic measurement of biparietal diameter and femur length performed on the day of serum sampling, menstrual dating agreeing within 10 days of ultrasonographic dating, and gestational age between 15 and 19 completed weeks. When these criteria were met in more than 200 samples for a given gestational week, every other sample was chosen. Maternal age and ultrasonographic measurements, as well as assurance that the infant was phenotypically normal, were provided by the referring physician. Both maternal serum AFP and hCG concentrations were measured by enzymelinked immunosorbent assay with commercially available kits (Abbott Laboratories, Abbott Park, Ill.). Determinations were made in duplicate and the coefficient of variation was required to be < 10%. For convenience in comparison, results for maternal serum AFP were expressed as international units per ml and maternal serum hCG was expressed as international units per liter. Samples were grouped according to completed gestational weeks and weekly medians for maternal serum AFP, and maternal serum hCG, and biparietal diameter/femur length ratios were determined. Smoothed medians were then calculated by weighted linear regression. Adjustment of maternal serum AFP and maternal serum hCG for variation in maternal weight was by regression of the nonadjusted multiples of the median with reported maternal weight. All analyses were accomplished with the SAS statistical packages (SAS Institute, Cary, N. C.).
1797
Table I. Smoothed weekly medians for possible screening parameters Maternal
Week
n
serum AFP
15 16 17
76 146 130 120
24.1 29.4 34.7 40.4
18
Maternal
serum keG
BPD/FL
25.0 21.0 17.0 12.7
1.73 1.66 1.60 1.53
BPD/ FL, Fetal biparietal diameter/femur length ratio.
Table II. Relationship between predictors of Down syndrome Variable 1
Maternal Maternal Maternal Maternal Maternal
serum serum serum serum serum
Variable 2
heG AFP heG AFP heG
r
p
-0.04 0.37 BPD/FL BPD/FL 0.06 0.16 0.05 0.22 Age 0.03 0.45 Age Maternal serum AFP 0.10 0.02
BPD/FL, Fetal biparietal diameter/femur length ratio; Age, maternal age; T, Spearman Rank correlation coefficient; p, probability.
the median closely approximated gaussian distribution with a Shapiro-Wilk statistic value of 0.98, whereas maternal serum AFP achieved a value of 0.96 (values may range from 0 to 1.0; 1.0 indicates completely normal distribution). The distribution of biparietal diameter/femur length with a value of 0.88 differed more from a completely normal distribution value of 0.88, but conformed sufficiently to allow clinical usage. Comment
Results
There was no correlation between maternal age and any of the three variables, either as uncorrected values or after transformation. Initial analysis confirmed the expected correlations between gestational age and maternal serum AFP (r = 0.23), maternal serum hCG (r = - 0.25), and biparietal diameterlfemur length ratio (r = - 0.33). Smoothed weekly medians for each variable were generated (Table I) and values were transformed into multiples of the median. This transformation nullified an association with gestational age. However, a significant positive correlation remained between maternal serum hCG and maternal serum AFP (r = 0.15). Because maternal serum AFP (r = - 0.18) and maternal serum hCG (r = - 0.23) were both varied inversely with maternal weight, both were subsequently expressed as weight-adjusted multiples of the median. After this transformation, there was no longer a correlation with maternal weight, but a small (r = 0.10) but statistically significant correlation between maternal serum AFP and maternal serum hCG remained (Table II). Logarithms of maternal serum hCG multiples of
The traditional method of patient selection for Down syndrome testing by amniocentesis has been maternal age. Unfortunately, with use of this criterion alone only 20% to 30% of women with a fetus with Down syndrome are offered amniocentesis.' After the initial report that women carrying fetuses with Down syndrome had lower than average maternal serum AFP values, t this observation was confirmed and tables allowing the use of combined maternal age and maternal serum AFP concentration to estimate the risk of Down syndrome were published. 5 Several recent repons 6 . 7 have confirmed the value of combining maternal age with maternal serum AFP values to select a population of women younger than the traditional cutoff age with a risk similar to that of older women. Unfortunately, even with the a~junctive, use of maternal serum AFP detection of a significantly increased proportion of fetuses with Down syndrome requires a high false-positive rate. s The potential of using a combination of noninvasive measures to increase the sensitivity or decrease the false positivity of Down syndrome screening is of great cur-
1798 O'Brien et al.
rent interest. Of the possible candidates for such a role, the biparietal diameter/femur length ratio and maternal serum hCG appear to be the most promising. Inclusion of the biparietal diameter/femur length ratio is unlikely to result in a significant increase in cost, because most programs require ultrasonographic confirmation of dating before final recommendation for amniocentesis. Unlike maternal serum AFP, which is measured primarily to screen for open neural tube defects, determination of maternal serum hGC would increase screening costs. Currently, however, it appears to be the most likely candidate as a second screen since it performs more efficiently than measurement of unconjugated estriol at similar cost. In this study we demonstrated that, after adjustment for gestational age and maternal weight, maternal serum AFP, maternal serum hCG, and biparietal diameter/femur length ratio are sufficiently independent from each other and from maternal age that they may be combined in a bayesian estimate of relative and final risk. The correlation between maternal serum AFP and maternal serum hCG found in our sample before adjustment for maternal weight (0.15) is similar to the value (0.14) reported by Wald et a1. 9 in samples for which maternal weight was not available. However, maternal weight was only partially responsible for the corE relation and it is probable that there is an underlying biologic association. In contrast, the lack 6f association between biparietal diameter/femur length ratio and either of the biochemical measures assures that the combination of retarded femur growth, increased production of hCG, and deficient production of AFP seen in Down syndrome does not occur in chromosomally normal fetuses. Although these results are encouraging, care is required before the widespread introduction of maternal serum hCG and biparietal diameter / femur length ratio for Down syndrome screening. First, estimation of relative risk on the basis of the distributions of normal and Down syndrome pregnancies is relatively straightforward, but there is some concern that the distribution of maternal serum AFP values in pregnancies associated with Down syndrome may be dependent on maternal age or on geographic site. IO The distribution of maternal serum hCG in pregnancies associated with Down syndrome has only been determined in a small number of cases, and similar concerns must be considered. Second, the values for the biparietal diameter/femur length ratios noted in this study are similar to those reported from Boston, but they differ from those generated in New Haven" with intermediate results reported from Pittsburgh. '2 Although it is possible
December 1990 Am J Obstet Cynecol
that these differences are based on population, it is more likely that they are related to technique. 2 Unlike maternal serum AFP or maternal serum hCG determinations, over which laboratories can exert quality control, most centers rely on ultrasonographic values generated by the referring physician. Significant variations in ultrasonographic technique among participating physicians may obscure the value of biparietal diameter/femur length ratio on screening. Finally, the cost effectiveness and goals of the screening program must be evaluated. Should maternal serum hCG be determined for all subjects or only those found at increased risk by evaluation of maternal age and maternal serum AFP? We have shown that the combination of maternal age, maternal serum AFP, maternal serum hCG, and biparietal diameterlfemur length ratio in the prediction of Down syndrome risk may be feasible. The proper implementation of such a program requires further study. REFERENCES 1. Bogart MH, Pandian MR,]ones OW. Abnormal maternal serum chorionic gonadotropin levels in pregnancies with fetal chromosome abnormalities. Prenat Diagn 1987; 7: 623-30. 2. Lockwood C, Benacerraf B, Krinsky A, et al. A sonographic screening method for Down syndrome. AM] OBSTET GYNECOL 1987;157:803-8. 3. Adams M, Erikson], Layder P, Oakley GP. Down's syndrome: recent trend in the United States. ]AMA 1981; 246:758-60. 4. Merkatz lR, Nitowski HM, Macri ]N, Johnson WE. An association between low maternal serum alphafetoprotein and fetal chromosomal abnormalities. AM] OBSTET GYNECOL 1984; 148:886-91. 5. Cuckle HS, Wald N], Lindenbaum RH. Maternal serum alpha-fetoprotein measurement: a screening test for Down syndrome. Lancet 1984; 1:926-9. 6. DiMaio MS, Baumgarten A, Greenstein RM, Saal HM, Mahoney MJ. Screening for fetal Down's syndrome in pregnancy by measuring maternal serum alphafetoprotein levels. N Engl] Med 1987;317:342-6. 7. Palomaki GE, William], Haddow]E. Combining maternal serum alpha-fetoprotein measurements and age to screen for fetal Down syndrome in pregnant women. AM ] OBSTET GYNECOL 1989;160:575-81. 8. Spencer K, Carpenter P. Screening for Down's syndrome using alpha-fetoprotein: a retrospective study indicating caution. Br Med] 1985;290: 1940-3. 9. Wald N], Cuckle HS, Densem ]W, et al. Maternal serum screening for Down's syndrome in early pregnancy. Br Med] 1988;297:883-7. 10. Hook EB. Variability in predicted rates of Down syndrome associated with elevated maternal serum alphafetoprotein levels in older women. Am] Hum Genet 1988;43:160-4. 11. Lockwood C, Benacerraf B, Krinsky A, et al. A sonographic screening method for Down syndrome. AM] OBSTET GYNECOL 1987; 157:803-8. 12. Hill LM, Guzick D, Belfar HL, Hixson], Rivello D, Rusnak ]. The current role of sonography in the detection of Down syndrome. Obstet Gynecol 1989;74:620-3.