| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
|
by the American Institute of Ultrasound in Medicine J Ultrasound Med 23:1613-1618 • 0278-4297 Reproducibility of the Fetal Nasal Bone Length MeasurementDepartments of Obstetrics and Gynecology (M.N.B., J.M.G.v.V.) and Clinical Epidemiology and Biostatistics (J.W.R.T.), Vrije Universiteit Medical Center, Amsterdam, the Netherlands. Address correspondence and reprint requests to Mireille N. Bekker, MD, Vrije Universiteit Medical Center, Stafgebouw L-348, de Boelelaan 1117, 1081 HV Amsterdam, the Netherlands. E-mail: mn.bekker{at}vumc.nl.
Objective. To investigate the reproducibility of the fetal nasal bone length measurement in the first trimester. Methods. In this prospective study, crown-rump length, nuchal translucency, and nasal bone length were measured in 90 singleton pregnancies between 11 and 14 weeks of pregnancy. The study was divided into 2 equal substudies. Three investigators measured the nasal bone in study I. After an intensive 3-month training program, the same 3 investigators measured the nasal bone in study II. Measurement of the nasal bone was standardized. All investigators were blinded and measured the nasal bone 3 times. The intraclass correlation coefficient (intra-CC) and interclass correlation coefficient (inter-CC) were calculated. Results. Nasal bone length measurement was successfully performed in all cases. The median nasal bone lengths were 2.3 (range, 1.5–3.2), 2.6 (range, 1.4–4.2), and 2.9 (range, 2.1–3.8) mm between 77 and 83, 84 and 90, and 91 and 98 days’ gestational age, respectively; median crown-rump length was 63.0 (range, 45.0–83.3) mm; and median nuchal translucency measurement was 1.3 (range, 0.9–2.5) mm. The correlation between the 3 observers was poor (inter-CC, 0.32) in the first study. After the training program, the interobserver correlation improved but was still moderate (inter-CC, 0.64). In both studies the interobserver variability did not differ between 11, 12, and 13 weeks’ gestational age. There was no relationship between the mean nasal bone length and interobserver difference in both studies. The intraobserver variability was good for each investigator in both studies (intra-CC: study I, 0.93, 0.95, and 0.97; study II, 0.98, 0.97, and 0.97, respectively). Conclusions. This study shows that the reproducibility of the fetal nasal bone length measurement in the first trimester is inadequate.
Key Words: Down syndrome • nasal bone length • prenatal • reproducibility • sonography Abbreviations: CRL, crown-rump length • inter-CC, interclass correlation coefficient • intra-CC, intraclass correlation coefficient • NT, nuchal translucency
In the last few decades, different screening strategies for Down syndrome have been suggested. First-trimester screening for Down syndrome is currently based on maternal age, nuchal translucency (NT) measurement, and maternal serum biochemical screening. This test has sensitivity of 90% for a false-positive rate of 5%.1 If screening reveals a high risk for Down syndrome, an invasive test will be offered. Invasive tests have a risk of miscarriage. Recently, the fetal nasal bone has been proposed as a novel sonographic marker for Down syndrome.2 One of the common characteristics of the Down syndrome genotype is a flat profile with a small nose.3 Postmortem radiologic and histologic studies of fetuses with Down syndrome document hypoplasia or delayed ossification of the nasal bone.4,5 Recent studies suggest that this hypoplasia can be evaluated with sonography and appears as an absence or shortened (<10th percentile) fetal nasal bone.6–12 For example, Cicero et al1 found an absent nasal bone in 69% of fetuses with Down syndrome and in 0.4% of fetuses with normal karyotypes at 11 to 14 weeks’ gestation. It was estimated that screening for trisomy 21 by combining maternal age, NT, maternal serum biochemical screening, and examination of the fetal nasal bone could increase the detection rate to 97% for a false-positive rate of 5%.1 An improvement in first-trimester screening for Down syndrome could result in a major reduction in the need for invasive testing and, therefore, a decrease in the number of miscarriages. Before fetal nasal bone assessment by sonography is incorporated into first-trimester screening for Down syndrome, its reproducibility should be evaluated extensively. In this prospective study, we investigated the reproducibility of the fetal nasal bone length measurement at 11 to 14 weeks’ gestation.
Women visiting our hospital for first-trimester screening for Down syndrome were asked to participate in the study. Exclusion criteria were twin pregnancy and unknown last menstrual period and cycle length. All patients received written information and gave informed consent. The Medical Ethical Committee of the Vrije Universiteit University Medical Center approved the study. A total of 90 patients were included in the study. The study contained 2 substudies (studies I and II), each consisting of 45 patients. The first study was the measurement of the nasal bone length before an extensive training program, and the second study was performed after the training program. Each study was divided into 3 equal groups with gestational ages of 11, 12, and 13 weeks, respectively. In study I, 3 investigators measured the length of the nasal bone in 45 fetuses. All 3 investigators were skilled in first-trimester sonography and were already experienced in nasal bone measurement before starting with the first study. Each fetus was examined for crown-rump length (CRL) and NT by the first investigator. Nuchal translucency measurement was performed according to the guidelines of the Fetal Medical Foundation (London, England).13 After this examination, the 3 investigators separately evaluated the fetal profile and measured the length of the fetal nasal bone according to the following protocol:
This protocol was based on previous studies by Cicero et al.2 Images of the examinations were stored digitally, videotaped, or both. Sonographic examinations were performed transabdominally with a 2- to 4-MHz probe (HDI 5000; Philips Medical Systems, Bothell, WA). In cases of inadequate fetal position, a 4- to 8-MHz transvaginal probe (HDI 5000) was used. Each investigator registered difficulties during the investigation. All 3 investigators were blinded to the results of the other investigators. Each investigator measured the nasal bone 3 times after relocating the bone. The second study was performed after a comprehensive training program. The training program consisted of a 3-month period of interactive training sessions in which videotapes, digitally stored images, and real-time investigations were discussed. In the second study, the same 3 investigators measured the fetal nasal bone in another 45 fetuses according to a protocol similar to that used in study I. The within- and between-observer variability was assessed by calculation of the intraclass correlation coefficients (intra-CCs) and interclass correlation coefficient (inter-CCs). Interclass correlation coefficients were calculated for all 3 examiners together and for each examiner pair. Bland-Altman plots (difference between 2 paired measurements plotted against the average between measurements) were constructed. Follow-up was obtained in each case by questionnaire, telephone contact, or karyotype results whenever chorionic villous biopsy or amniocentesis was performed.
Nasal bone length measurement was successfully performed in all 90 cases. In 2 cases, transvaginal sonographic examination was required because of the fetal position. All other measurements were performed transabdominally. The median nasal bone lengths were 2.3 (range, 1.5–3.2), 2.6 (range, 1.4–4.2), and 2.9 (range, 2.1–3.8) mm between 77 and 83, 84 and 90, and 91 and 98 days’ gestational age, respectively; the median maternal age was 35 (range, 24–40) years; the median CRL was 63.0 (range, 45.0–83.3) mm; and the median NT measurement was 1.3 (range, 0.9–2.5) mm. Reported difficulties during the measurement included maternal adipositas (n = 3), obtaining the correct angle of insonation because of an unfavorable fetal position (n = 6), and placing the calipers accurately at the end of the nasal bone (n = 12). All 90 fetuses had a normal karyotype antenatally or a normal phenotype at birth.
The intraclass correlation for 1 observer was excellent for all 3 observers both before and after the training program (Table 1
In both studies, the interobserver variability did not differ at 11, 12, and 13 weeks’ gestational age as well for all 3 examiners together as between different pairs of examiners. Table 3  shows the inter-CCs at different weeks’ gestational age between the most concordant pair of examiners in both studies. Analysis in the other examiner pairs led to the same findings and is therefore not shown.
There was no relationship between the mean nasal bone length and interobserver difference both before and after intensive training. This is shown in the Bland-Altman plots in Figures 2 and 3. Before the training program, the mean difference between observers was 0.073 mm, with a large SD of 0.41 mm. After the training program, some improvement was observed. The mean difference was 0.034 mm; however, the SD was still large: 0.31 mm.
Sonographic examination of the fetal nasal bone has recently been proposed as an additional tool in the first-trimester risk assessment for Down syndrome.2 Valuable screening should be both feasible and reproducible, in addition to having high specificity and sensitivity. These criteria should be critically evaluated before the nasal bone examination can be routinely used in a clinical setting. This study shows that the fetal nasal bone length measurement in the first trimester is only fairly reproducible. The reproducibility improved after extensive training but remained inadequate, with the highest inter-CC of 0.71 for the best examiner pair. The different gestational age groups showed similar inter-CCs in both studies. Therefore, we conclude that the repeatability does not improve with increasing gestational age. Two other studies of the variability in fetal nasal bone assessment have been performed. Senat et al14 found moderate reproducibility of nasal bone identification on sonography in a retrospective study of video loops of 30 seconds’ duration. Consequently, in a prospective setting, even higher interoperator disagreement can be expected. A study by Kanellopoulos et al6 showed acceptable intraobserver and interobserver variation in fetal nasal bone length measurement in the first trimester. Linear regression analysis showed that the length of the nasal bone increased with CRL. On the basis of this regression analysis, they concluded that the detection and measurement of the nasal bone becomes easier as gestation advances. Nevertheless, our Bland-Altman plots clearly show that the interobserver variation did not improve with a longer nasal bone length; that is, the absolute difference between observers did not change with increasing nasal bone length. As a consequence, with increasing nasal bone length, the difference between measurements represents a smaller percentage of the total length. Furthermore, our Bland-Altman plots show that, although intensive training leads to less variation between measurements, inadequate repeatability between observers still remains. Cicero et al15 studied the number of scans necessary for an experienced sonographer to be competent in examining the fetal nasal bone in the first trimester. They concluded that, on average, 80 scans were required to perform an accurate evaluation of the presence of the nasal bone. Our examiners already had extensive experience in nasal bone examination before starting with the first study. The intraoperator variability was good, and a correct image of the nasal bone could be obtained in all cases. The interobserver variation, however, was poor in the first study and became moderate after the second study. The learning curve for accurate length measurement seems to be long, even for highly skilled sonographers. Sonographic evaluation of the fetal nasal bone in screening for trisomy 21 was introduced as identification of the nasal bone. Recent studies, however, have shown that hypoplasia rather than absence is the case in trisomy 21.4–6 Minderer et al16 compared sonographic with pathomorphologic findings in fetuses with trisomy 21. A retrospective review of sonographic images identified nasal bones in 5 of 6 cases in which they were initially reported as being absent on sonography. The nasal bones were found to be hypoplastic. Therefore, an accurate nasal length measurement with good reproducibility has become more important than nasal bone identification. Several difficulties in nasal bone length measurement should be mentioned. First of all, the nasal bone length in the first trimester is relatively small compared with other fetal biometric measurements. In the correct magnification, each increment between calipers has a minimal distance of 0.1 mm. However, each 0.1 mm is a large percentage of the nasal bone length (mean percentages at 11, 12, and 13 weeks’ gestational age, respectively: 4.4% [range, 3.2%–6.7%], 3.8% [range, 2.5%–7.1%], and 3.5% [range, 2.6%–4.7%]). A major problem, together with the small size, is caliper placement because of lateral scattering and new ossification of the nasal bone. The nasal bone length is measured from end to end. In the proposed angle of insonation by Cicero et al,2 the ultrasound beam had a 90° angle with the nasal bone and was best visualized. At this angle, the lateral scatter is difficult to distinguish from the actual nasal bone, which results in a false increase in the length. Changing the angle can reduce the lateral scattering. This, however, is difficult to standardize. Another problem with caliper placement is the echo density of the fetal nasal bone. The nasal bone ossification develops on both sides of the cartilaginous septum from 10 weeks’ gestational age.17–19 New ossification of the nasal bone appears as less echogenic extensions at the sides of the bone. It has been proposed that only the echogenic part of the nasal bone should be measured. Nevertheless, it remains difficult to distinguish lateral scattering from newly ossified bone. Also, in cases of trisomy 21, the nasal bone can be less echogenic because of the delayed ossification.4 Furthermore, the nasal bones develop bilaterally from their own ossification center, which is independent of the contralateral bone.19,20 Differences in timing of development and ossification between both nasal bones could influence the sonographic length measurement. This could also explain the large interobserver variability observed in this study. Additionally, Tuxen et al20 histologically examined 33 fetuses with Down syndrome. Two fetuses had unilateral absence of the nasal bone. To our knowledge, the difference in length between both nasal bones has never been studied in fetuses with normal karyotypes. The role of 3-dimensional sonography in examining the nasal bones should be explored. Three-dimensional sonography has the advantage of a multiplanar view.21,22 This technique may lead to an adequate examination of both nasal bones separately. Especially, the maximal mode technique has proved to be of additional value in examining the fetal skeleton.21 This modality makes it possible to exclusively select the densest structures, such as bones, and could therefore make a more accurate measurement possible. Difficulties with the 2-dimensional examination, such as lateral scattering, could be reduced. In conclusion, this study shows that the reproducibility of fetal nasal bone length measurement is inadequate. At this point, the nasal bone investigation is not ready for routine use in a clinical setting. Efforts should be made for good standardization, training, and quality control. Further studies are needed to validate the nasal bone length measurement in screening for Down syndrome in a general population.
Received June 30, 2004, from the Departments of Obstetrics and Gynecology (M.N.B., J.M.G.v.V.) and Clinical Epidemiology and Biostatistics (J.W.R.T.), Vrije Universiteit Medical Center, Amsterdam, the Netherlands. Revision requested August 2, 2004. Revised manuscript accepted for publication August 26, 2004. We thank A. Druijff, M. Engels, A. Maat, and M. Rekoert-Hollander for assistance.
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Top
Abstract
Introduction
