Down syndrome occurs in approximately one in 800 live births, with typical findings of this syndrome including mental retardation, abnormal facies, heart defects, and short stature. Using current techniques, 24–93% of Down syndrome fetuses are detected by high-resolution ultrasound during the second trimester.1,2 Much effort has been directed at increasing this detection rate by incorporating specific ultrasound findings or markers into formulas used to adjust a woman's a priori risk.1–6 A few currently used markers include a thickened nuchal fold, echogenic bowel, echogenic intracardiac foci, renal pyelectasis, and shortened humeri and femurs.1–16
Many investigators have evaluated the yield of fetal long bone biometry as a screening tool for Down syndrome. A measured:expected femur length (FL) ratio of 0.91 or less is considered a risk factor for Down syndrome, where the expected FL is calculated based on the fetal biparietal diameter (BPD). Several investigators used this ratio to adjust a woman's risk by giving a point in a sonographic scoring system (where a score of two or greater is considered abnormal) or by increasing a woman's a priori risk for Down syndrome by 2.2.3–5 Lockwood et al6 reported that using a BPD:FL ratio greater than 1.5 standard deviations (SD) above the mean for unaffected controls was a sensitive method for detecting Down syndrome. This method of screening for Down syndrome appears to have an increased sensitivity over using a measured:expected FL ratio.7 Drugan et al8 summarized many ultrasound findings and reported a relative risk (RR) of 2.7 for Down syndrome based on the finding of short FL.
Recently, investigators have shown that the FLs in second-trimester fetuses are shorter in Asians and Hispanics compared with blacks and whites (Kovac CM, Brown J, Hume RF, Apodaca CC, Pierce B, Napolitano P, et al. Maternal ethnicity and second-trimester fetal femur length variation. Abstract to be presented at the 45th annual convention of the American Institute of Ultrasound in Medicine, 2001). This finding indicates that certain patient characteristics can influence the fetal FL calculations and falsely elevate a woman's risk. The goal of the current study was to determine whether the measured to expected FL ratio or the BPD:FL ratio is influenced by gestational age or maternal height.
Materials and Methods
Madigan Army Medical Center is a tertiary care facility and referral center for military beneficiaries in the Pacific Northwest and Alaska. Three full-time ultrasonographers with 28 years of combined experience perform all ultrasound evaluations in the antepartum diagnostic center in conjunction with perinatalogy consultation. Ultrasound examinations are done using one of three machines, an ATL HDI 3000 (ATL Ultrasound, Bothell, WA) and two Acuson 128 XPs (Acuson Corporation, Mountain View, CA). Ultrasound examinations are recorded on an AEGIS Database Client (Acuson Corporation) recording package and a printout is made for the patient's record. Maternal-fetal medicine staff review all ultrasound examinations.
We prospectively evaluated fetal FLs for all patients who had second-trimester ultrasound between 15 0/7 weeks' gestation and 19 6/7 weeks' gestation from January 1997 through May 2000. The patient population included high-risk and low-risk women, with high risk defined as women 35 years of age or older, with abnormal maternal serum analyte screening results or with fetal anomalies. Patients with incomplete records, incomplete fetal measurement data, or fetal death at the time of examination were excluded. The dating criteria used to calculate gestational age were based on the patients' last menstrual period if the average ultrasound age was within 8 days of the menstrual dates. Otherwise the average ultrasound age was used for pregnancy dating. If a patient had an earlier ultrasound that confirmed or changed her dates, that was used as her dating criteria. All patients who had an incomplete fetal anatomic survey were reexamined at a later gestational age, but patients were included only once for study purposes.
Four formulas were used to calculate the expected FLs based on BPD, two from published literature,9,10 one derived from the patient population overall, and one derived from the patient population specific to each gestational age. The FLs were plotted against the BPD, and the best-fit line was used for calculation. For all gestational ages, a polynomial equation was generated by multiple linear regression analysis secondary to the nonlinear relationship between the BPD and FL (Figure 1). For the individual gestational ages, simple regression analysis was used to calculate the best-fit line, as the relationship appeared linear. A measured:expected FL ratio of 0.91 or less was considered abnormal for all formulas used.
All patients who had ultrasound examination at 15 weeks' gestation and again at 18 or 19 weeks' gestation were identified. The mean measured:expected FL ratio was calculated using both previously described formulas for the 15-week ultrasound and compared with the mean ratio at the 18- or 19-week ultrasound. Regression analysis was used to determine a correlation between maternal height and fetal FLs for each gestational age.
The mean maternal height and the mean BPD:FL ratio were calculated for each gestational age. Using a BPD:FL value greater than 1.5 SD above the mean as abnormal, we compared the number of abnormal values for women shorter than one SD below the mean height with the number of abnormal values for women taller than one SD above the mean height. The relative risks and 95% confidence intervals for having a BPD:FL ratio greater than 1.5 SD above the mean were calculated.
The computer program Statview for Windows (Abacus Concepts, Inc., Berkeley, CA) was used for all calculations. χ2 analysis was used to assess differences in patient ethnicity, and an unpaired t test was used to assess differences in age or height. A paired t test was used to assess differences in the measured:expected FL ratio for the 15-week examination compared with the same patient's 18- or 19-week examination. P < .05 was considered statistically significant for all analyses.
A fetal anatomic survey was completed on 1650 patients between 15 0/7 weeks and 19 6/7 weeks: 31 patients from 15 0/7 to 15 6/7 weeks; 67 patients from 16 0/7 to 16 6/7 weeks; 172 patients from 17 0/7 to 17 6/7 weeks; 877 patients from 18 0/7 to 18 6/7 weeks; and 503 patients from 19 0/7 to 19 6/7 weeks.
Patient demographic characteristics are listed in Table 1. There was no significant difference between groups with respect to age, ethnicity, or height.
For our population, the formulas used to generate the expected femur lengths are as follows: all gestational ages, expected FL = −3.222 + 2.194(BPD) − .182(BPD2); 15 weeks, expected FL = −0.169 + 0.62(BPD); 16 weeks, expected FL = 0.726 + 0.395(BPD); 17 weeks, expected FL = 0.787 + 0.435(BPD); 18 weeks, expected FL = 0.932 + 0.431(BPD); and 19 weeks, expected FL = 1.211 + 0.406(BPD).
Using those formulas and previously described formulas, the percentage of women with abnormal measured:expected FL ratios of 0.91 or less decreased with increasing gestational age (Table 2). For each formula used, there was a significant increase in RR of having an abnormal ratio at 15 to 16 weeks' gestation compared with 18 to 19 weeks' gestation (Table 3).
Sixteen of 31 women who had an ultrasound examination at 15 weeks' gestation also had a repeat examination at 18 or 19 weeks' gestation. An increase in measured:expected FL ratio was still found at the later gestational age using both previously published formulas (mean measured:expected FL = 0.940 [at 15 weeks] compared with 0.988 [at 18–19 weeks]10; mean measured:expected FL = 1.022 [at 15 weeks] compared with 1.058 [at 18–19 weeks]9; P < .05 for both).
There was a statistically significant correlation between maternal height and fetal FL at 18 and 19 weeks' gestation (P < .05), which did not exist at 15, 16, or 17 weeks' gestation (P > .05).
For 18- and 19-week gestations, women shorter than one SD below the mean had a higher risk of abnormal BPD:FL ratio compared with women taller than one SD above the mean (21 of 153 for women shorter than one SD compared with 12 of 208 for women taller than one SD (RR 2.38; 95% CI 1.21, 4.69).
Biparietal diameter:FL ratios decreased with advancing gestation (P < .001; BPD:FL = 2.618 − 0.06(gestational age); R2 = .212). The mean (± one SD) BPD:FL ratios for each gestational age are as follows: 15 weeks, 1.777 ± 1.34; 16 weeks, 1.666 ± 0.128; 17 weeks, 1.571 ± 0.114; 18 weeks, 1.532 ± 0.096; and 19 weeks, 1.480 ± 0.092. There was no correlation between BPD and maternal height or measured:expected FLs and maternal height overall or for each gestational age (P > .05).
Current methods of detecting fetuses with Down syndrome based on abnormal FLs are influenced by gestational age and maternal height. Women who have genetic ultrasound examinations at earlier gestational ages have higher risk of abnormal measured:expected FL ratios compared with women at later gestational ages. This higher risk might be explained by the fact that current methods for calculating the expected FLs, as based on the BPD, overestimate the FL at the early gestational ages. We developed a polynomial equation to calculate the expected FL based on BPD from 15 0/7 weeks to 19 6/7 weeks, as well as a gestational age-specific formula. Both formulas also had a higher risk at earlier gestational ages for an abnormal ratio. Of the four formulas presented, the gestational age–specific formula had a slightly higher abnormal rate compared with that described by Nyberg et al,9 while having a lower RR for an abnormal ratio at the earlier gestational ages. The formula described by Benacerraf et al10 has the lowest RR overall for the earlier gestational ages but had a very high abnormal rate. This will lead to increased sensitivity for Down syndrome detection based on a sonographic scoring system, but it will also lead to increased false-positive results. Our data suggest that current formulas used to calculate measured to expected FLs falsely elevate a woman's risk of fetal Down syndrome when calculated at earlier gestational ages.
The prevalence of Down syndrome is dependent on gestational age, and more fetuses are expected to have Down syndrome at 15 weeks compared with 18 or 19 weeks' gestation. The spontaneous loss of fetuses with Down syndrome between those gestational ages might explain some of the increase in measured:expected FLs at the later gestational ages. However, the prevalence of Down syndrome, even in high-risk populations, is low, as is the risk of fetal loss specifically between the gestational ages of 16 and 18 weeks. On the basis of these facts, the impact of the possible loss of Down syndrome fetuses is a small contribution to the increased rate of abnormal ratios at the earlier gestational ages. To evaluate the effect of the loss of Down syndrome or other anomalous fetuses, we compared the measured:expected FL of 16 of the 31 fetuses who were included in the 15-week ultrasound group to their measured:expected FL at 18 or 19 weeks' gestation. There was still a higher measured:expected FL ratio at the later gestational age using both previously published formulas, which discounts that the increase in measured:expected FLs was due to a loss of anomalous fetuses between these gestational ages.
Another possible explanation of the higher rate of abnormal measured:expected FLs at the earlier gestational ages is that, at the earlier gestational ages with smaller fetuses, the risk of operator error increases. A small difference in measurement of the BPD or FL will have a much greater effect on the measured:expected FL ratio at 15 weeks' gestation compared with 19 weeks' gestation. The fetal biometry, while otherwise adequate, is too imprecise to have reliable estimates of measured:expected femur lengths for smaller fetuses. This measurement should not be used for Down syndrome screening at the early gestational ages because of the high rate of abnormal values. A reasonable gestational age to implement this ratio is 17 weeks' gestation and beyond, when the positive rate is 5% or less9 (Table 2).
Screening for Down syndrome using an abnormal BPD:FL ratio, as defined by a ratio over 1.5 SD above the mean for each gestational age might be useful at the earlier gestational ages of 15 to 17 weeks. At and beyond 18 weeks' gestation, genetic factors begin to influence the fetal FL, leading to a two-times higher risk of an abnormal ratio for women shorter than 5 feet 1 inch, compared with women taller than 5 feet 7 inches. One reason for the lack of correlation between maternal height and fetal femur length at the earlier gestational ages could be that there is not enough statistical power to detect a correlation, as far fewer patients had ultrasound examination at those gestational ages.
The results obtained in this study should be able to be reproduced and generalizable to other populations, as accurate femur length measurements are fairly easy to obtain. However, ethnic variation might explain the poor correlation of our best-fit line compared with published data (R2 = .667 compared with R2 = .909),9 as our military population might be more ethnically diverse.
The most accurate method of screening for Down syndrome using fetal FL measurements remains to be determined. Further investigation is needed with greater numbers of patients at the earlier gestational ages, as well as consideration of ethnicity, to better predict the expected FLs based on the BPD. Providers who counsel patients concerning their risk of fetal Down syndrome must take into account the influence of gestational age on the measured-to-expected FL ratio, as well as the influence of maternal height on the BPD:FL ratio at the later gestational ages.
1. Nyberg DA, Luthy DA, Resta RG, Nyberg BC, Williams MA. Age-adjusted ultrasound risk assessment for fetal Down's syndrome during the second trimester: Description of the method and analysis of 142 cases. Ultrasound Obstet Gynecol 1998;12:8–14.
2. Vintzileos AM, Campbell WA, Rodis JF, Guzman ER, Smulian JC, Knuppel RA. The use of second-trimester genetic sonogram in guiding clinical management of patients at increased risk for fetal trisomy 21. Obstet Gynecol 1996;87:1–8.
3. Benacerraf BR, Nadel A, Bromley B. Identification of second trimester fetuses with autosomal trisomy by use of a sonographic scoring index. Radiology 1994;193:135–40.
4. Benacerraf BR, Neuberg D, Bromley B, Figoletto F. Sonographic scoring index for prenatal detection of chromosomal abnormalities. J Ultrasound Med 1992;11:449–58.
5. Snijders RJM, Nicolaides KH, eds. Ultrasound markers for fetal chromosomal defects. Carnforth, United Kingdom: Parthenon Publishing, 1996.
6. Lockwood C, Benacerraf B, Krinsky A, Blakemore K, Belanger K, Mahoney M, et al. A sonographic screening method for Down syndrome. Am J Obstet Gynecol 1987;157:803–8.
7. Dicke JM, Gray DL, Songster GS, Crane JP. Fetal biometry as a screening tool for the detection of chromosomally abnormal pregnancies. Obstet Gynecol 1989;74:726–9.
8. Drugan A, Johnson MP, Evans MI. Ultrasound screening for chromosome anomalies. Am J Med Genet 2000;90:98–107.
9. Nyberg D, Resta R, Luthy D, Hickok D, Williams M. Humerus and femur length shortening in the detection of Down's syndrome. Am J Obstet Gynecol 1993;168:534–8.
10. Benacerraf BR, Cnaan A, Gelman R, Laboda L, Frigoletto F. Can sonographers reliably identify anatomy features associated with Down syndrome? Radiology 1989;173:377–9.
11. Vintzileos AM, Campbell WA, Guzman ER, Smulian JC, McLean DA, Ananth CV. Second trimester markers for detection of trisomy 21: Which markers are best? Obstet Gynecol 1997;89:941–4.
12. Sepulveda WH, Cullen S, Nicolaideis P, Hollingsworth J, Fisk NM. Echogenic foci in the fetal heart: A marker of chromosomal abnormality. Br J Obstet Gynaecol 1995;102:490–2.
13. Benacerraf BR, Neuberg D, Frigoletto FD. Humeral shortening in second-trimester fetuses with Down syndrome. Obstet Gynecol 1991;77:223–7.
14. Bromley B, Doubilet P, Frigoletto FD, Krauss C, Estroff JA, Benacerraf BR. Is fetal hyperechogenic bowel on second-trimester sonogram an indication for amniocentesis? Obstet Gynecol 1994;83:647–51.
15. Nicolaides KH, Shawwa L, Brizot M, Snijders RJ. Ultrasonographically detectable markers of fetal chromosomal defects. Ultrasound Obstet Gynecol 1993;3:56–9.
© 2001 The American College of Obstetricians and Gynecologists
16. Benacerraf BR, Mandell J, Estrff JA, Harlow B, Frigoletto FD. Fetal pyelectasis: A possible association with Down syndrome. Obstet Gynecol 1990;76:58–60.