Best Second Trimester Sonographic Markers for the Detection of Trisomy 21

Best Second Trimester Sonographic Markers for the Detection of Trisomy 21 Patrizia Vergani, MD, Anna Locatelli, MD, Maria Giovanna Piccoli, MD, Patrizia Ceruti, MD, Eloisa Mariani, MD, John C. Pezzullo, PhD, Alessandro Ghidini, MD We analyzed all genetic sonograms obtained during a 6 year period to establish the independent ability of the following sonographic markers of aneuploidy in the diagnosis of trisomy 21: structural anomalies, cardiac abnormalities, nuchal fold thickness of 6 mm or greater, bowel echogenicity, choroid plexus cysts, and renal pyelectasis. With the exception of bowel echogenicity and choroid plexus cysts, the sonographic markers were more common in trisomy 21 than euploid fetuses (all P < 0.001). Logistic regression analysis demonstrated that cardiac anomalies (odds ratio = 255; 95% confidence interval, 25, 2592), other structural anomalies (odds ratio = 25; 95% confidence interval, 6, 97), and nuchal fold thickness of 6 mm S econd trimester genetic amniocentesis carries an unavoidable risk of procedurerelated loss of normal fetuses. Several ABBREVIATIONS NFT, Nuchal fold thickness; FL, Femur length; HL, Humerus length; BPD, Biparietal diameter; SD, Standard deviation; OR, Odds ratio; CI, Confidence interval; ANOVA, Analysis of variance Received February 1, 1999, from the Divisione di Ostetricia e Ginecologia, Istituto di Scienze Biomediche San Gerardo, Monza, Italy (P.V., A.L., M.G.P., P.C., E.M.); and the Department of Obstetrics and Gynecology, Georgetown University Medical Center, Washington, DC (J.C.P., A.G.). Revised manuscript accepted for publication April 4, 1999. Address correspondence and reprint requests to Patrizia Vergani, MD, Department of Obstetrics and Gynecology, ISBM San Gerardo, via Solferino 16, 20052 Monza, Italy. or greater (odds ratio = 13; 95% confidence interval, 3, 50) were the only independent predictors of trisomy 21. The false-positive rate and sensitivity were 5.3% (48 of 898) and 59.2% (13 of 22), respectively, when any of the sonographic markers significant at univariate analysis was considered, and 3.1% (28 of 898) and 54.5% (12 of 22), respectively, when any of the predictors at multivariate analysis was present. Because a considerable overlap of sonographic markers exists among trisomy 21 fetuses, use of those that are not independent predictors leads to an increase in false-positive rate without a gain in sensitivity. KEY WORDS: Trisomy 21; Fetus, echocardiography; Fetus, anomalies; Down syndrome. authors have proposed serologic tests, such as maternal serum triple screening, during the early second trimester to adjust the maternal age related risk of fetal aneuploidy. 1,2 Similarly, ultrasonography offers the opportunity to better select candidates for prenatal diagnosis by using a series of markers that are present more frequently in aneuploid than euploid fetuses. 3 5 The term genetic sonogram has thus been coined. 6 As the number of proposed markers grows, the false-positive rate of the genetic sonogram inevitably increases. The repercussions of this approach are particularly worrisome in women without prior risk for trisomy 21, in whom the higher false-positive rates translate into a greater proportion of procedurerelated losses of euploid fetuses. To obviate this problem, multivariate analysis offers the optimal mode for selecting the markers with independent predictive ability that should be included in a genetic sonogram. 1999 by the American Institute of Ultrasound in Medicine J Ultrasound Med 18:469 473, 1999 0278-4297/99/$3.50

470 SONOGRAPHIC MARKERS FOR TRISOMY 21 J Ultrasound Med 18:469 473, 1999 In the current prospective cohort study, we evaluated a series of eight sonographic markers of trisomy 21 in an attempt to establish which ones are independent predictors. MATERIALS AND METHODS During a 6 year period (January 1, 1990, to December 31, 1996), all women with singleton fetuses receiving genetic counseling because of maternal age 35 years or older at delivery underwent a genetic sonogram during the early second trimester (14 to 22 weeks gestation). Maternal serum biochemical screening was not used in the study population. All ultrasonographic examinations were performed by six physicians with expertise in prenatal diagnosis and without prior knowledge of fetal karyotype. After excluding voluntary terminations of pregnancy, for which cases karyotype analysis was not available (n = 3), in utero deaths without available karyotype (n = 5), preterm neonatal death without karyotype (n = 1), and chromosome anomalies other than trisomy 21 (n = 11), 920 cases were available for analysis. The second trimester ultrasonographic examination (Ultramark 9, Advanced Technology Laboratories, Bothell, WA) included evaluation of fetal biometry and anatomy for detection of structural anomalies. Particular attention was also paid to the following ultrasonographic markers of aneuploidy: fourchamber view of the heart and outflow tracts with the use of color flow mapping, 7 NFT of 6 mm or greater, renal pyelectasis (anteroposterior diameter of renal pelvis greater than 4 mm), choroid plexus cysts, and hyperechogenic bowel (echogenicity similar to that of bones). Nuchal cystic hygromas were considered structural anomalies. Consenting women with a normal genetic sonogram obtained earlier than 18 weeks gestation were invited to return at 20 weeks for a more complete evaluation of the heart. Cytogenetic examinations were performed either by amniocentesis in consenting patients or at birth in cases in which it was clinically indicated if amniocentesis had been declined. Neonatal follow-up evaluation was available in 100% of cases. Statistical Analysis Regression analysis was performed for FL and HL measurements as functions of the BPD. Based on the regression equations, expected values of FL and HL for a given BPD were calculated, and the ratios of observed to expected values were compared between euploid and trisomy 21 fetuses. Univariate analysis was performed using one-way ANOVA for continuous variables and chi-square or Fisher s exact test for dichotomous variables. Logistic regression analysis was employed to correct for confounding variables. A P value less than 0.05 was considered significant. RESULTS Of the 920 women enrolled in the study, 22 had fetuses with trisomy 21. Mean gestational age at ultrasonography was 17.0 weeks (SD ± 1.7; range, 14 to 22 weeks). Mean maternal age was 38.4 years (SD ± 2.2; range, 35 to 47 years). Amniocentesis was performed in 311 of 898 (35%) euploid and 13 of 22 (59%) aneuploid cases (P < 0.001). Nine women with aneuploid fetuses declined prenatal cytogenetic testing despite the presence of fetal abnormalities in three of them. We had no false-negative prenatal diagnoses of major structural abnormalities or cardiac anomalies among trisomy 21 cases. The follow-up ultrasonographic examination at 20 weeks did not lead to detection of additional cardiac malformations. The ultrasonographic markers found in fetuses with trisomy 21 are shown in Table 1, whereas the results of univariate analysis are displayed in Table 2. Mean ± SD NFT values were 3.2 ± 0.9 mm and 5.0 ± 1.7 mm in euploid and aneuploid fetuses, respectively (P < 0.001). The presence of any of the sonographic markers that were significantly different between euploid and trisomy 21 fetuses at univariate analysis yielded a false-positive rate of 5.3% (48 of 898) with a sensitivity of 59.2% (13 of 22). FL was available in 21 and HL in seven Down syndrome fetuses. Mean ± SD observed versus expected values of FL and HL were not significantly different between euploid and aneuploid fetuses (Table 2). Gestational age was significantly different between fetuses with normal versus abnormal heart at ultrasonography (17.0 ± 1.7 weeks versus 19.3 ± 2.7 weeks, P < 0.001). Similarly, a correlation was present between gestational age and NFT (R = 0.39, P < 0.001). Logistic regression analysis demonstrated that after controlling for gestational age, cardiac anomalies (OR, 255; 95% CI, 25, 2592), other structural anomalies (OR, 25; 95% CI, 6, 97), and NFT of 6 mm or greater (OR, 13; 95% CI, 3, 50) were the only independent predictors of trisomy 21. When any of these markers was present, the sensitivity for the diagnosis of trisomy 21 was 54.5% (12 of 22), and the false-positive rate was only 3.1% (28 of 898). To test if a negative genetic ultrasonogram was independent of maternal age, we performed a twoway ANOVA for maternal age, karyotype results, and presence of any ultrasonographic markers of

J Ultrasound Med 18:469 473, 1999 VERGANI ET AL 471 Table 1: Ultrasonographic Findings in Trisomy 21 Cases. Maternal Gestational Structural Cardiac NFT Case Age (yr) Age (wk) Anomalies Anomalies Pyelectasis (mm) 1 41 14 None None No 6 2 44 15 None None Yes 5 3 43 16 None None No 3 4 40 16 None AVSD No 5 5 37 16 None None No 4 6 40 16 None None No 3 7 40 16 Ventriculomegaly None No 4 8 42 16 None None No 6 9 41 17 None None No 2 10 42 17 None None No 5 11 45 18 None AVSD No 4 12 40 18 None None No 3 13 36 20 None None No 4 14 35 20 Duodenal atresia None Yes 3 15 37 20 None None No 4 16 39 21 Ventriculomegaly None No 7 17 38 21 None None No 4 18 36 21 None None No 8 19 40 22 None AVSD Yes 4 20 37 22 None Ebstein anomaly Yes 7 21 41 22 Duodenal atresia AVSD No 6 22 43 22 Hydrocephaly None No 5 AVSD, Complete atrioventricular septal defect. aneuploidy. Whereas maternal age was significantly different between euploid and aneuploid cases, as expected (P = 0.04), it was not different between cases with any versus none of the independent sonographic predictors of trisomy 21 (P = 0.22). DISCUSSION We found that half of the sonographic markers for trisomy 21 that we examined were not significantly different between euploid and trisomy 21 fetuses in our population. Moreover, at logistic regression analysis an additional marker (pyelectasis) was found not to be an independent predictor of trisomy 21. In other words, although pyelectasis is significantly more common among trisomy 21 than euploid fetuses, as it has been reported previously, 8 most cases of pyelectasis in trisomy 21 fetuses occur among those with other markers, such as increased NFT or structural or cardiac anomalies. As a corollary, detection of pyelectasis in a fetus should prompt a more thorough evaluation in search of the markers that are independent predictors of trisomy 21 rather than providing counseling directed at calculating the pyelectasis-associated risk of Down syndrome. During the last 15 years, the number of genetic sonographic markers proposed has increased progressively. As a consequence, the complexity of genetic sonograms has grown, and the type of markers used varies from center to center. The components of each genetic sonogram are often chosen by prioritizing the ones that are more easily obtained and more familiar to the operator. As our analysis shows, inclusion of markers that are not independent predictors leads to an increase of 41.5% in the false-positive rate, with little improvement in sensitivity. A genetic sonogram should include visualization of the independent predictors, which are, on the basis of our results, NFT and cardiac outflow tracts, in addition to conducting the usual search for structural abnormalities. At variance with results of other series, 9 11 fetal HL was not a predictor of trisomy 21 in our study, possibly owing to the small number of cases in which this parameter was included. Three studies in addition to ours utilized multivariate analysis to evaluate the independent predictors of trisomy 21 (Table 3). 9 11 Two of them originated from the same center and had considerable overlap in the study periods. 10,11 NFT is the only marker that has consistently been found to have independent predictive ability in all series. It should

472 SONOGRAPHIC MARKERS FOR TRISOMY 21 J Ultrasound Med 18:469 473, 1999 Table 2: Univariate Analysis of the Sonographic Markers for Trisomy 21* Sonographic Marker Euploid (n = 898) Trisomy 21 (n = 22) P Value Cardiac anomalies 3 (0.3%) 5 (22.7%) < 0.001 Structural anomalies 11 (1.2%) 5 (22.7%) < 0.001 Nuchal fold thickness 6 mm 16 (1.8%) 6 (27.3%) < 0.001 Pyelectasis 18 (2.0%) 4 (18.2%) 0.001 Isolated choroid plexus cysts 24 (2.6%) 1 (4.5%) 0.45 Bowel echogenicity 7 (0.8%) 0 0.85 Humerus length 1.00 ± 0.10 1.04 ± 0.07 0.38 Femur length 1.00 ± 0.10 0.96 ± 0.07 0.08 *Values given as number (%) or mean ± SD. Observed versus expected values based on biparietal diameter. be noted that the value of a detailed cardiac examination cannot be assessed from this review because only one study specified that cardiac outflow tracts were evaluated in addition to the four-chamber view of the heart. 9 Cardiac malformations were excluded from the analysis in two of the studies, 10,11 whereas the third study included cardiac anomalies among other structural abnormalities. 9 Whether visualization of the outflow tracts adds predictive ability to that offered by the four-chamber view alone cannot be assessed from our study. In our investigation, 22.7% of trisomy 21 fetuses had cardiac anomalies in the four-chamber view of the heart and outflow tracts with the use of color flow mapping. This rate compares favorably with those reported by Benacerraf and coworkers (21%) 12 and by DeVore and Alfi (18%). 7 A potential limitation of cardiac anomalies as a marker of trisomy 21 is that an adequate view of the heart may not be possible until 18 weeks or later. In support of the reliability of cardiac evaluation at low gestational ages comes the observation of De Vore and colleagues, who, using color flow mapping, correctly identified cardiac anomalies in 10 of 12 Down syndrome fetuses at or below 18 weeks of gestational age. 7 If fetal cardiac anatomy cannot be adequately assessed before 18 weeks gestation, and the patient declines a follow-up examination, counseling should take into account all the other sonographic markers that are independent predictors of trisomy 21. Multivariate analysis permits a more accurate counseling in the presence of multiple sonographic markers. The cumulative odds for trisomy 21 can be calculated by multiplying the individual odds ratios of the markers with independent predictive ability. In conclusion, multivariate analysis of the sonographic markers of trisomy 21 allows identification of the optimal components of a genetic sonogram. This may lead over time to a standardization of the contents of such examinations among different centers and populations, a shorter duration of the examination with no loss of sensitivity, and a decrease in the false-positive rate, with its associated risks for euploid fetuses undergoing invasive genetic testing. Table 3: Comparison of Sonographic Markers Found to be Independent Predictors of Trisomy 21 Markers Vintzileos et al (1997) 9 Deren et al (1998) 10 Bahado-Singh et al (1998) 11 Present Series Nuchal fold thickening Significant Significant Significant Significant Humerus shortening Significant Significant Significant NS Femur shortening NS NS NS NS Pyelectasis Significant NS NS Hyperechoic bowel NS Significant Significant NS Choroid plexus cysts NS NS Hypoplastic middle phalanx NS Significant Significant Sandal gap* NS Two-vessel umbilical cord NS NS Tibia shortening NS Ulna shortening NS NS, Not significant at multivariate analysis. The empty cells refer to markers not considered in the studies. Structural malformations were controlled for. *Increased space between first and second toe.

J Ultrasound Med 18:469 473, 1999 VERGANI ET AL 473 REFERENCES 1. Haddow JE, Palomaki GE, Knight GJ, et al: Reducing the need for amniocentesis in women 35 years of age or older with serum markers for screening. N Engl J Med 330:1114, 1994 2. Wenstrom KD, Desai R, Owen J, et al: Comparison of multiple-marker screening with amniocentesis for the detection of fetal aneuploidy in women > 35 years old. Am J Obstet Gynecol 173:1287, 1995 3. Vintzileos AM, Egan JFX: Adjusting the risk for trisomy 21 on the basis of second trimester ultrasonography. Am J Obstet Gynecol 172:837, 1995 4. Nadel AS, Bromley B, Frigoletto FD, et al: Can the presumed risk of autosomal trisomy be decreased in fetuses of older women following a normal sonogram? J Ultrasound Med 14:297, 1995 5. Bahado-Singh RO, Özgür D, Tan A, et al: Ultrasonographically adjusted midtrimester risk of trisomy 21 and significant chromosomal defects in advanced maternal age. Am J Obstet Gynecol 175:1563, 1996 6. Vintzileos AM, Campbell WA, Rodis JF, et al: The use of second-trimester genetic sonogram in guiding clinical management of patients at increased risk for fetal trisomy 21. Obstet Gynecol 87:948, 1996 7. DeVore GR, Alfi O: The use of color Doppler ultrasound to identify fetuses at increased risk for trisomy 21: An alternative for high risk patients who decline genetic amniocentesis. Obstet Gynecol 85:378, 1995 8. Benacerraf BR, Mandell J, Estroff JA, et al: Fetal pyelectasis: A possible association with Down syndrome. Obstet Gynecol 76:58, 1990 9. Vintzileos AM, Campbell WA, Guzman ER, et al: Secondtrimester ultrasound markers for detection of trisomy 21: Which markers are best? Obstet Gynecol 89:941, 1997 10. Deren O, Mahoney MJ, Copel JA, et al: Subtle ultrasonographic anomalies: Do they improve the Down syndrome detection rate? Am J Obstet Gynecol 178:441, 1998 11. Bahado-Singh R, Deren O, U Oz, et al: An alternative for women initially declining genetic amniocentesis: Individual Down syndrome odds on basis of maternal age and multiple ultrasonographic markers. Am J Obstet Gynecol 179:514, 1998 12. Benacerraf BR, Neuberg D, Bromley B, et al: Sonographic scoring index for prenatal detection of chromosomal abnormalities. J Ultrasound Med 11:449, 1992