Evaluation of multiple-marker screening for Down syndrome in a statewide population

Evaluation of multiple-marker screening for Down syndrome in a statewide population Katharine D. Wenstrom, MD," Roger A. Williamson, MD," Stanley S. Grant, RN," Jerry D. Hudson, BS,b and Jane P. Getchell, DPHb

Iowa City, Iowa OBJECTIVE: Our purpose was to evaluate our experience with a statewide, multiple-marker Down syndrome screening program. STUDY DESIGN: The results of 18,712 screening tests performed from July 1,1991, to Oct. 31,1992, were reviewed. Amniocentesis and aneuploidy detection rates were compared with the experience of a previous year (1989-1990) in which material serum a-fetoprotein was used for detection of Down syndrome. RESULTS: Positive screening tests (Down syndrome risk :::: 1/190) occurred in 665 of 18,712 (3.5%) patients; 516 of 665 (78%) patients accepted amniocentesis. Fifteen aneuploidies were identified: 12 trisomy 21, one trisomy 18, one trisomy 13, and one 48,XXXY. The overall detection rate was one in 34 amniocenteses performed; for trisomy 21 it was one in 43. In a previous year in which maternal serum a-fetoprotein alone was used, 3.6% had positive screening tests (Down syndrome risk:::: 270); the detection rate for all aneuploidies was one in 57 amniocenteses, and for trisomy 21 it was one in 114. The expanded maternal serum a-fetoprotein test was well accepted by clinicians, with 36% of gravid state residents undergoing screening. CONCLUSION: The multiple marker test is a good screening tool and is superior to material serum a-fetoprotein alone. (AM J OBSTET GVNECOL 1993;169:793-7.)

Key words: Down syndrome, multiple-marker screening, maternal serum a-fetoprotein

The association between low maternal serum a-fetoprotein (AFP) levels and fetal aneuploidy was first reported by Merkatz et al. I Maternal serum AFP was then rapidly evaluated as a screening tool to detect pregnancies at risk for fetal Down syndrome. 2 Subsequently a large collaborative trial in which 77,273 women < 35 years old were screened in the second trimester confirmed that 25% of fetuses with Down syndrome could be identified by means of a combination of maternal serum AFP and maternal age." The continuing search for additional markers to improve predictive accuracy has resulted in identification of two hormones whose levels are changed by fetal aneuploidy. Both maternal serum unconjugated estriol 4 and human chorionic gonadotropin 5 levels are altered in pregnancies complicated by fetal Down syndrome. The combination of maternal serum AFP, estriol, human chorionic gonadotropin, and maternal age (the multiple-marker test) has been evaluated retrospecFrom the Department of Obstetrics and Gynecology" and the University Hygienic Laboratory,' University of Iowa Hospital. Presented at the Thirteenth Annual Meeting of the Society of Perinatal Obstetricians, San Francisco, California, February 8-13,1993. Reprint requests: Katharine D. Wenstrom, MD, Department of Obstetrics and Gynecology, Old Hillman Building, Room 457, 618 20th St. 5., Birmingham, AL 35233-7333. Copyright © 1993 by Mosby-Year Book, Ine. 0002-9378/93 $1.00 + .20 6/6/49176

tively on stored serum, and it is estimated to identify > 60% of affected pregnancies. 6 A large prospective study by the Foundation for Blood Research 7 has now confirmed the predicted efficacy of multiple-marker screening in detecting fetal Down syndrome. We used this same protocol to initiate a state-wide multiple-marker screening program in Iowa. We now report our experience with 18,712 multiple-marker tests performed from July 1, 1991, through Oct. 31, 1992. Material and methods The State of Iowa Maternal Serum AFP Screening Program was initiated in 1984 and since then has provided centralized laboratory and clinical genetics services to all gravid state residents. In 1989 a pilot study was conducted with the support of the state health department to establish means, medians, SDs, and correlation coefficients for estriol and human chorionic gonadotropin in the Iowa population. On July 1, 1990, a multiple-marker screening program (called the Expanded MSAFP Screening Program) was begun. By July I, 1991, all laboratory and clinical aspects of the program were functioning efficiently and accurately. Follow-up of screen-positive patients has been good (93%) since that time. All serum samples were drawn between 15 and 20

793

794 Wenstrom et al.

weeks' gestation, and analyzed with commercially available radioimmunoassay kits (Amersham, Arlington Heights, Ill.). Maternal serum AFP values were corrected for maternal weight, race, and presence of insulin-dependent diabetes. Down syndrome risk was determined with all three analyte levels (evaluated with a trivariate gaussian algorithm to correct for slight pairwise correlation of the three analytes) and maternal age. The test was considered screen positive if the Down syndrome risk was ~ 1 : 190. We changed from our old maternal serum AFP risk cutoff of 1: 270 to maintain our false-positive rate (unnecessary amniocentesis rate) at an acceptably low level while still identifying the maximum number of Down syndrome cases. The 1 : 190 cut off was chosen because it was predicted to combine a reasonable detection rate (60%) with a low falsepositive rate (4.5%). With the old risk cutoff of 1 : 270 we estimated that our false-positive rate would have increased to 7.2%,8 resulting in the performance of > 200 additional amniocenteses to detect only one more case of Down syndrome. The physicians of all patients with a positive initial screen were contacted, and they verified the estimated gestational age by local ultrasonography. Fetal biparietal diameter measurements were used to confirm age because they are unchanged by Down syndrome. 9 If the initial estimated gestational age was determined to be incorrect, the Down syndrome risk was either recalculated on the basis of the revised dates (if the sample was drawn at ~ 15 weeks's gestation) or a second sample was obtained (if the sample was drawn at < 15 weeks' gestation). All patients ultimately determined to have a Down syndrome risk ~ 1 : 190 were counseled and offered genetic amniocentesis for diagnosis. Amniocentesis was also offered to patients ~ 35 years old without a screening test or regardless of test results. Older patients undergoing elective amniocentesis were not included in the analysis. The majority of screen-positive patients accepted amniocentesis; follow-up of those refusing definitive testing was obtained by contacting the referral physician directly. Complete follow-up of all screen-negative patients is not currently possible. An estimate of the completeness of ascertainment was made by comparing the number of Down syndrome cases identified with the multiple-marker screen to the number theoretically expected in our screened population. The expected frequency of midtrimester Down syndrome was calculated using the maternal ages of the screened population and established age-specific Down syndrome incidence. lo We also compared our experience with multiple-marker screening to that of a previous year (1989 to 1990) with maternal serum AFP alone. Statistical analysis was performed with Mantel-Haenszel X2 with p :5 0.05 considered significant.

October 1993 Am J Obstet Gynecol

Results

FromJuly 1, 1991, to Oct. 31, 1992, 18,712 screening tests were performed (Fig. 1). This number represents roughly 36% of all gravid state residents. Positive screening tests (Down syndrome risk ~ 1 : 190) occurred in 665 (3.5%). Amniocentesis was accepted in 516 (78%). Fifteen aneuploidies were identified: 12 trisomy 21, one trisomy 18, one trisomy 13, and one 48,XXXY. The Down syndrome detection rate was one in 43 amniocenteses performed; the detection rate for all aneuploidies was one in 34. These figures compare favorably with a previous year's experience with maternal serum AFP alone, in which 3.6% of all screening test results were positive (Down syndrome risk ~ 1: 270) and the detection rate per amniocentesis for one in 114 for Down syndrome and one in 57 for all aneuploidies (Table I). The screen-positive patients were grouped according to maternal age and level of risk and compared to evaluate the decision to accept amniocentesis (Fig. 2). Patients with Down syndrome risk > 1 : 50 were significantly more likely to undergo amniocentesis than were their lower risk cohorts (p = 0.049, Mantel-Haenszel odds ratio 1.594, 95% confidence intervals 1.002 to 2.538). Patients ~ 35 years old were more likely to refuse amniocentesis, regardless of degree of risk (p = 0.000, Mantel-Haenszel odds ratio 0.257, 95% confidence interval 0.177 to 0.373). This was not surprising, because the patients > 35 years old who underwent multiple-marker screening had already refused amniocentesis for advanced maternal age. The 15 aneuploidies identified were then reviewed with respect to maternal age and level of expanded maternal serum AFP screen risk (Table 11). More cases were identified among the high-risk patients (risk > 1 : 50) and among patients ~ 35 years old, although the numbers were too small to achieve significance. Follow-up of the 149 screen-positive patients who refused amniocentesis revealed one live-born infant with trisomy 21. In addition, five unkaryotyped intrauterine fetal deaths occurred. Twenty-eight patients are currently undelivered and 10 have been lost to follow-up. Using established age-specific Down syndrome incidences and the maternal ages of the screened population, we determined that 27 cases of Down syndrome were expected. Thirteen cases have been definitively identified in the screen-positive group (48%). Because one third of second-trimester losses involve fetal aneuploidy and because 29% of all continuing Down syndrome pregnancies result in fetal death in the second or third trimester, I I it is likely that at least one or two of the screen-positive intrauterine fetal deaths were trisomic. It is also possible that one of the currently undelivered screen-positive patients has a Down syn-

Wenstrom et al.

Volume \69, Number 4 Am J Obstet Gynecol

C

795

18,712 women screened

J. 665 amnios recommended 3.5%

.J. 516 women accepted 78%

l •

12 cases DS

+

1 case trisomy 18 1 case trisomy 13 1 case 48 XXXV

Down Syndrome: 1 casei 43 procedures

All: 1 case/34 procedures

DS = Down syndrome Fig. 1. Triple marker screening in Iowa, July I, 1991, to Oct. 31, 1992.

Table I. Comparison of expanded screen with maternal serum AFP alone

Risk cutoff Positive screens Detection rate: Down syndrome All

Expanded maternal serum AFP screen (1991)

Maternal serum AFP

;:: 1:190 3.5%

;:: 1:270 3.6%

1/43 1/34

1/114 1/57

drome fetus. Our detection rate may therefore be as high as 15 in 27 or 56%.

Comment Chromosome abnormalities characterize 60% of all spontaneous abortions, 0.5% of term births, and 6% to 7% of all stillbirths. 12 Because of the devastating effects of aneuploidy, the relative frequency of occurrence, and the availability of definitive diagnostic tests, prenatal diagnosis of fetal karyotypic abnormalities has been the subject of much research. After the report of Merkatz et al. 1 in 1984 linking low maternal serum AFP levels and fetal aneuploidy, the possibility of using maternal serum AFP to screen the low-risk population aroused great interest. Both retrospective studies 13 and large prospective trials 2 • 3 confirmed that maternal serum AFP levels combined with maternal age could detect 25% of all cases of fetal Down syndrome. Research then focused on ways to improve screening efficacy while decreasing the number of amniocenteses performed. Because a-fetoprotein is produced in the fetal liver, the search for other diagnostically useful hormones initially focused on compounds requiring the liver for synthesis. Thus it

(1989)

was determined that unconjugated estriol is on average 27% lower in Down syndrome pregnancies" Evaluation of placentally derived proteins revealed that human chorionic gonadotropin levels are roughly twice those of normal gestations. 5 All three analyte alterations are consistent with Down syndrome fetuses and placentas behaving as if they are 2 to 3 weeks less mature than their normal counterparts. Both estriol and human chorionic gonadotropin have been evaluated retrospectively and prospectively and have been shown to add substantially to the accuracy of maternal serum AFP screening. It has been estimated that approximately 60% of all cases can be detected when all three markers plus maternal age are used to calculate Down syndrome risk. 6 Haddow et aJ.7 have now prospectively evaluated the multiple-marker screen in 25,207 women in the New England area. They reported a detection rate of 58% while increasing the diagnostic efficiency of amniocentesis to one case of Down syndrome identified per 38 procedures performed. In addition to Down syndrome, they identified cases of sex chromosome aneuploidy, trisomy 13, and familial balanced chromosomal rear-

796

Wenstrom et al.

Octobcr 199 ~ Am J Obstct " Yll ccol

100

**

80 0/0

o >1 :50 o 1:50-1:99

60 40

Ell 1:100-1:149 • 1: 150- 1:190

20 0 <25

25-29

30-34

~35*

All Ages

·p=
Fig. 2. Percentage of patient s undergoing amniocentesis according to risk level.

Table 11. Aneuploidies ide ntified according to risk level

Maternal age ()'IJ

> 1:50

<25 25-29 30-34

I Down syndrome

~35

All ages (No. of cases per No. of am niocenteses)

5 Down syndrome I Trisomy 13 7/ 154

1:50-!- 99

I: 100-1: 149

I: /50-1: 190

3 Down syndrome

I Trisomy 18 I Down syndrome

I Down syndrome

4/1 92

2/164

2/155

146,XXXY

rangements, for a n overall detection frequency of one in 28 procedures. The multiple-ma rker screen has also been tes ted prospectively in the sta te of Washington by C h e ng e t a!. 11 In an analysis of 7718 sam ples 6% were screen positive at a risk cut otf of 2= I : 195 . Ninety-one p e rce lll of Down syndrome pregnancies were identified by a n abnormal screen. Evaluation of these remarkably good results is complica ted by the fact that the actual number of Down syndrome cases in the screened popula tion was fully 47% greater than expected (22 cases identified vs 15 cases expected). In addi tion, 18% of the 22 cases resulted from translocations, a situation usually responsible for only 3% to 5% of Down syndrome. Data from this unusual p a tient population may not extrapolate to other geographic areas or screening centers . Our experience is consistent with earlier predictions of the efficacy of the multiple-marker scree n and reiterates the prospective experience of the Foundation for Blood Research. The d e tection rate per amnioce ntesis performed was substantially better than that usin g m aternal serum AFP levels a lo ne. The test was re la tive ly well accepted by pregnant women and their physicians,

Down syndrome

All ages (No. of rases pn' No. of . amniocenteses) 21100 21165 4/23 1 7/665 15/665

with 36% of pregn a nt state residents undergoing screemng. In summary, the multiple marker test is a good screening tool and is superio r to maternal serum AFP levels alone . It allows improved deteCtion of Down syndrome and appears to detect a neuploidies other than trisomy 2 1. REFERENCES I. Merkatz JR. Nitowsky HM, Macri IN, Johnson WE. An

2.

3.

4.

5.

association between low maternal serum a-fetoprotein and fetal chromosomal abnormalities. AM .I OBsn:T G\,!'H:C:O I. 1984; 148:886-94. DiMaio MS, Baum garte n A, Greenstein RM, Saal HM , Mahoney MJ. Screening for fetal Down syndrome in pregnancy by measuring maternal serum a-fetoprotein levels. N Engl J Med 1987;3 17:342-6. New England Regional Genetics Group Pren ata l Collaborative Study of Down Syndrome Screening. Combining maternal serum a-fetoprotein measurements and age to screen for Down syndrome in pregnant women under age 35 . A.M J OIJSTET CI':\rcoL 1989; 160:575-8 1. Canick .lA, Knight GJ, Palomaki CE, Haddow .l E, Cuckle HS, Walk NJ. Low second-trimester maternal serum unconjugated oestriol in pregnancies with Down syndrome. Br J Obstet Gynaeco l 1988;95:330-3. Bogart MH, Pandian MR, J ones OW. Abnormal maternal serum chorionic gonadotropin levels in pregnancies with

Volume 169, Number 4 Am J Obstet Gynecol

6. 7. 8. 9.

10.

fetal chromosome abnormalities. Prenat Diagn lY87;7: 623-30. Wald N], Cuckle HS, Densem .lW, et al. Maternal serum screening for Down syndrome in early pregnancy. BM] 1988;297 :883-7. Haddow ]E, Palomaki GE, Knight G.l, et al. Prenatal screening for Down syndrome with use of maternal serum markers. N Engl .l Med 1992;327:588-93. Canick]A, Knight GJ. Multiple-marker screening for fetal Down syndrome. Contrib Obstet Gynecol Tech 1992:2592. Cuckle HS, Wald N.l. The effect of estimating gestational age by ultrasound cephalometry on the sensitivity of a-fetoprotein screening for Down's syndrome. Br ] Ob stet Gynaecol 1987;94:274-6. Cuckle HS, Wald N], Thompson SG. Estimating a woman's risk of having a pregnancy associated with Down syndrome using her age and serum a-fetoprotein level. Br ] Obstet Gynaecol 1987;94:38-42.

Wenstrom et al.

797

11. Hook EB, Topol BB, Cross PK. The natural history of cytogenetically abnormal fetuses detected at mid trimester amniocentesis which are not terminated e1ectively: new data and estimates of the excess and relative risk of late fetal death associated with 47,t21 and some other abnormal karyotypes. Am .l Hum Genet 1989;45:855-61. 12. Emery AEH, Rimoin DL. Nature and incidence of genetic disease. In: Emery AEH, Rimoin DL. Principles and practice of medical genetics, vol 1. 2nd ed. New York: Churchill Livingstone, 1990:3. 13. Doran TA, Cadesky K, Wong PY, Mastrogiacomo C, Capello T. Maternal serum a-fetoprotein and fetal autosomal trisomies. AM .l OBSTE'I GYNECOL 1986;154:277-81. 14. Cheng EY, Luthy DA, Zebelman AM, WilIiams MA, Luppman RE, Hickok DE. A prospective evaluation of a secondtrimester screening test for fetal Down syndrome using maternal serum a-fetoprotein, hCG, and unconjugated estriol. Obstet Gynecol 1993;81 :72-7.

Bound volumes available to subscribers Bound volumes of the AMERICAN JOURNAL OF OBSTETRICS AND GYNECOLOGY are available to subscribers (only) for the 1993 issues from the Publisher, at a cost of $74.50 for domestic, $102.72 for Canada, and $97.50 for international for Vol. 168 (January-June) and Vol. 169 (J uly-December). Shipping charges are included. Each bound volume contains a subject and author index and all advertising is removed. Copies are shipped within 60 days after publication of the last issue in the volume. The binding is durable buckram with the JOURNAL name, volume number, and year stamped in gold on the spine. Payment must accompany all orders. Contact Mosby, Subscription Services, 11830 Westline Industrial Drive, St. Louis, MO 63146-3318, USA; phone (800) 453-4351 or (314) 453-4351. Subscriptions must be in force to qualify. Bound volumes are not available in place of a regular JOURNAL subscription.